510(k) Predicate Devices: How to Find, Evaluate, and Select the Right Predicate

Step-by-step guide to finding and selecting predicate devices for FDA 510(k) submissions — search strategies, substantial equivalence criteria, split predicates, and common selection mistakes.

Why Predicate Selection Is the Most Important Decision in Your 510(k)

Predicate selection is not a checkbox. It is the strategic foundation of your entire 510(k) submission. Every section of your filing — device description, intended use statement, performance testing, comparison tables, substantial equivalence argument — flows directly from which predicate you choose. Pick the wrong one and you face months of rework, additional information (AI) requests, or a Not Substantially Equivalent (NSE) determination that sends you back to square one.

The FDA reviews approximately 3,000 to 3,500 510(k) submissions per year. Among the most common reasons for rejection: inadequate substantial equivalence comparison and inappropriate predicate device selection. These are not documentation errors you can fix with a paragraph edit. They are strategic failures that require a fundamentally different approach.

This guide covers everything you need to know about finding, evaluating, and selecting predicate devices — from the legal definition and FDA database search techniques to split predicates, predicate chains, common mistakes, and a complete 7-step selection workflow.

Consider the real cost of getting this wrong. A company that selects an inappropriate predicate and receives an NSE determination faces not just the loss of the 510(k) user fee (over $26,000 for FY2026, standard rate) but also months of accumulated testing costs, staff time, and — critically — delayed market entry. For startups operating on venture timelines, an NSE can be existential. For larger companies, it disrupts product launch schedules and competitive positioning. The predicate selection decision deserves the same rigor you apply to your most important engineering decisions.

Related: For a complete walkthrough of the 510(k) submission process, including eSTAR formatting, user fees, and review timelines, see our Complete Guide to FDA 510(k) Submissions. For FDA device classification fundamentals, see our FDA Device Classification Guide.

What Is a Predicate Device?

A predicate device is a legally marketed device to which a new device is compared in a 510(k) submission to demonstrate substantial equivalence. The term comes from Section 510(k) of the Federal Food, Drug, and Cosmetic Act (FD&C Act) and the implementing regulations at 21 CFR Part 807.

Legal Definition

Under federal law, a predicate device is a legally marketed device that is either:

A pre-amendment device — a device that was legally marketed in the United States before May 28, 1976, when the Medical Device Amendments were enacted
A device cleared through the 510(k) pathway — a device that received a substantially equivalent (SE) determination from the FDA
A device classified through De Novo — a device that was authorized via the De Novo classification process under Section 513(f)(2) of the FD&C Act and is not exempt from premarket notification requirements

The critical word is "legally marketed." A device does not need to still be on the market today to serve as a predicate. It needs to have been legally marketed at some point. This distinction has important implications, which we will address in the sections on withdrawn and rescinded devices below.

What a Predicate Is NOT

A predicate is not:

A device that was never cleared or authorized by the FDA
A device that was removed from the market by FDA enforcement action (e.g., seized as adulterated or misbranded by judicial order)
A device classified through PMA (premarket approval) — PMA-approved devices cannot serve as predicates for 510(k) submissions
A device marketed solely under an IDE (Investigational Device Exemption) — investigational devices are not legally marketed

Key distinction: The FDA clears 510(k) devices and grants De Novo authorizations. It approves PMA devices. These are different legal statuses, and only 510(k)-cleared devices and De Novo-authorized devices (that are not exempt from 510(k)) can serve as predicates.

Types of Predicates

Not all predicates serve the same function in a 510(k) submission. Understanding the distinction between primary predicates, reference predicates, and the now-prohibited split predicates is essential.

Primary Predicate

The primary predicate is the single legally marketed device to which you claim substantial equivalence. Your SE determination lives or dies on this comparison.

Requirements for your primary predicate:

Same intended use as your device (or your intended use is a subset of the predicate's)
Same or comparable technological characteristics — and if different, you must demonstrate that the differences do not raise new questions of safety or effectiveness
Legally marketed status — confirmed through the FDA's 510(k) database

Every 510(k) submission must have one clearly identified primary predicate. The FDA evaluates substantial equivalence against this device.

Reference Predicate (Supporting Predicate)

A reference predicate is an additional legally marketed device you cite to support specific aspects of your SE argument. Reference predicates do not replace the primary predicate — they supplement the comparison.

Common uses for reference predicates:

Demonstrating that a specific material or component has been previously cleared for a similar clinical application
Showing that a particular technology (e.g., wireless connectivity, a sensor type, a software algorithm) has an established regulatory track record
Providing additional performance data context for a feature that differs from the primary predicate

Example: You are submitting a new wireless pulse oximeter. Your primary predicate is a recently cleared wireless pulse oximeter from a competitor (same intended use, same basic technology). You also cite a reference predicate — an older cleared device that uses the same novel optical sensor material — to demonstrate that this material has been previously cleared for contact with skin in a monitoring application.

Important: Each reference predicate cited in your submission must have the same intended use as your device. You cannot use a reference predicate to borrow intended use from a device in a different clinical application.

Split Predicates — What They Were, and Why FDA Banned Them

A "split predicate" was the practice of demonstrating substantial equivalence by combining two or more predicate devices that had different intended uses — claiming the intended use from one device and the technological characteristics from another device with a different intended use.

The FDA has prohibited split predicates. The agency's guidance on evaluating substantial equivalence makes clear that each predicate device cited in a 510(k) submission must share the same intended use as the subject device. You cannot "split" the SE determination by borrowing intended use from Device A and technology from Device B when those two devices serve different purposes.

Why did the FDA ban this practice? Because split predicates created a path for devices to reach the market without any single predicate truly matching the new device's combination of intended use and technology. This undermined the purpose of the 510(k) pathway — ensuring that new devices are as safe and effective as devices already on the market.

Predicate Strategy	Allowed?	Description
Single primary predicate	Yes	One predicate with same intended use and same/comparable technology
Primary + reference predicates (same intended use)	Yes	Multiple predicates, all sharing the same intended use
Split predicates (different intended uses)	No	Combining intended use from one device with technology from another that has a different intended use

If your device truly combines features from devices with different intended uses, the 510(k) pathway may not be appropriate. Consider whether a De Novo classification request is the right path. A Pre-Submission (Q-Sub) meeting with the FDA can help clarify the appropriate pathway.

How to Search for Predicate Devices: FDA Databases and Tools

Finding the right predicate requires a systematic search across multiple FDA databases. Below is a step-by-step approach using the publicly available tools.

Overview of FDA Databases for Predicate Research

Before diving into the step-by-step process, here is a summary of the databases you will use:

Database	URL	Purpose
Product Classification Database	accessdata.fda.gov/scripts/cdrh/cfdocs/cfPCD/classification.cfm	Identify product code, classification, and special controls
510(k) Premarket Notification Database	accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm	Search all 510(k) clearances — the primary predicate search tool
GUDID	accessgudid.nlm.nih.gov	Cross-reference commercial product names with 510(k) numbers
MAUDE	accessdata.fda.gov/scripts/cdrh/cfdocs/cfmaude/search.cfm	Adverse event reports for candidate predicates
Recall Database	accessdata.fda.gov/scripts/cdrh/cfdocs/cfres/res.cfm	Recall history for candidate predicates
De Novo Database	accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/denovo.cfm	De Novo authorizations that may serve as predicates

Step 1: Identify Your Product Code

Go to the FDA Product Classification Database: accessdata.fda.gov/scripts/cdrh/cfdocs/cfPCD/classification.cfm

Search by device name, description, or keyword. The product code is a three-letter code (e.g., "QKQ" for pulse oximeters, "DXN" for orthopedic screws, "QAS" for computerized electrocardiograph analysis) that determines:

Your device classification (Class I, II, or III)
The review panel and division that will handle your submission
Applicable special controls and guidance documents
Which 510(k)s to search for predicates

Tip: If your device could fit multiple product codes, identify all of them. Your predicate does not need to share the same product code as your device — but if it does not, you need to explain why the comparison is still valid. Using the same product code simplifies the review.

Step 2: Search the 510(k) Premarket Notification Database

Go to the FDA 510(k) Database: accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm

This is the primary tool for finding cleared devices. Search strategies:

Search by product code (most reliable):

Enter your three-letter product code in the Product Code field
Leave all other fields blank — the FDA search engine uses exact matching, so filling multiple fields simultaneously often returns zero results
Sort results by Decision Date (newest first) to find recent clearances

Search by applicant name (when you know the manufacturer):

If you know a specific competitor's device, search by their company name
This reveals all their cleared devices, including different product codes they may have used

Search by device name (use with caution):

Use a single descriptive keyword, not a full product name
The database matches exact text, so "glucose monitor" will not find "blood glucose monitoring system"
Try multiple keyword variations

Search by 510(k) number (when you have a specific clearance):

Enter the K-number (e.g., K212345) to pull up the exact clearance record
Useful when you already have candidates and want to confirm details

Step 3: Download and Review 510(k) Summary Documents

For each promising predicate candidate, click through to the clearance record and download the 510(k) Summary (if available). Not every clearance has a public summary — some submitters file a "510(k) Statement" instead, which contains minimal information.

A 510(k) Summary reveals:

The predicate's intended use / indications for use statement (critical for your comparison)
Predicate devices cited by that submitter (allows you to trace the predicate chain)
Testing performed — what bench testing, biocompatibility, electrical safety, software testing, and clinical data the FDA reviewed
Standards cited — which consensus standards (ISO, IEC, ASTM) were referenced
Basis for the SE determination — the FDA's reasoning for finding the device substantially equivalent

Pro tip: Prefer predicates that have a public Summary rather than a Statement. The Summary gives you a detailed roadmap of what the FDA found acceptable. A Statement tells you almost nothing.

Step 4: Cross-Reference with the GUDID

The Global Unique Device Identification Database (GUDID): accessgudid.nlm.nih.gov

GUDID provides:

Brand names and commercial product names linked to 510(k) numbers
Device identifiers (UDI/DI)
Company contact information
Device characteristics (single use, sterile, MRI compatible, etc.)

This is particularly useful when you know a competitor's commercial product name but not their 510(k) number. Search by brand name in GUDID, find the 510(k) number, then pull up the clearance details in the 510(k) database.

Step 5: Check the MAUDE Database for Adverse Events

The Manufacturer and User Facility Device Experience (MAUDE) Database: accessdata.fda.gov/scripts/cdrh/cfdocs/cfmaude/search.cfm

Before committing to a predicate, search MAUDE for adverse event reports (MDRs) associated with the candidate device. A predicate with a significant adverse event history is still legally valid, but:

It may invite additional scrutiny from reviewers
The FDA's draft "Best Practices" guidance recommends selecting predicates without unmitigated safety issues
If the predicate has known safety problems, reviewers may ask you to demonstrate that your device does not share those risks

For a detailed guide to the MAUDE database, see our FDA MAUDE Database Guide.

Step 6: Search the FDA Recall Database

The FDA Recall Database: accessdata.fda.gov/scripts/cdrh/cfdocs/cfres/res.cfm

Check whether any candidate predicate has been subject to a Class I, II, or III recall. Pay special attention to design-related recalls (as opposed to manufacturing or labeling recalls). The FDA's draft "Best Practices" guidance specifically recommends against selecting a predicate with an associated design-related recall.

A device with a recall history can still serve as a predicate. But you should:

Understand the root cause of the recall
Be prepared to explain why your device does not share the deficiency
Consider whether a different predicate without a recall history would strengthen your submission

Step 7: Use FOIA for Detailed Review Information

Freedom of Information Act (FOIA) requests: fda.gov/regulatory-information/freedom-information

For the most detailed view of how the FDA evaluated a predicate, submit a FOIA request for the review memorandum. The review memo contains the FDA reviewer's complete analysis — far more detailed than the public 510(k) Summary. It often includes:

Reviewer comments on testing adequacy
Questions raised during the review and how they were resolved
The specific rationale for the SE determination
Any concerns the reviewer noted but did not formally pursue

FOIA requests are free but may take weeks to months. Submit them early in your development process.

Search Strategy Tips

Start broad, then narrow. Begin with a product code search that returns all clearances in your device category. Scan the results to understand the competitive landscape, identify major manufacturers, and see the range of devices that have been cleared. Then narrow by reviewing individual summaries for the most promising candidates.

Search multiple product codes. If your device could reasonably fit under more than one product code, search all of them. Your ideal predicate may have been classified under a different code than you expected.

Look at your competitors' predicates. When you find a competitor's 510(k) clearance, download their summary and note which predicate they used. Their predicate may also be a good candidate for your submission. This technique — tracing the predicate chain of competitors — often surfaces predicates you would not find through keyword searching alone.

Track the regulatory landscape over time. If you are early in product development, set up periodic searches (monthly or quarterly) of the 510(k) database for your product code. New clearances may create better predicate options than what was available when you started development.

Use third-party tools cautiously. Several commercial platforms offer enhanced 510(k) database search capabilities with features like predicate chain visualization, AI-powered matching, and competitive intelligence dashboards. These can be valuable supplements, but always verify results against the FDA's official databases. Third-party tools may have data gaps or lag behind FDA updates.

Leverage the openFDA API for programmatic searches. The FDA provides a free, public API at open.fda.gov that gives programmatic access to the 510(k) clearance database, device classifications, adverse events (MAUDE), and recalls. The openFDA Device 510(k) endpoint contains premarket notification records from 1976 to the present and supports structured queries by product code, applicant, decision date, and other fields. If you are comfortable with API calls or scripting, the openFDA API allows you to pull bulk data, build custom filters, and automate periodic predicate landscape monitoring in ways the FDA's web-based search interface cannot. Several free open-source tools (such as the FDA 510(k) Predicate Finder on GitHub) are built on this API and offer a more user-friendly interface for querying clearance data.

Supplement 510(k) Summaries with additional sources. Do not rely solely on the information in a 510(k) Summary to evaluate a candidate predicate. Summaries vary widely in completeness and detail. Supplement your research with the predicate device's instructions for use (IFU), service manuals, the manufacturer's website, published clinical literature, and — where possible — hands-on evaluation of the device itself. If the predicate's public summary lacks critical information about materials, performance specifications, or testing, you may need to generate your own comparison data through bench testing or request the full review file through FOIA.

Evaluating Candidate Predicates: A Systematic Approach

Once your search yields a list of candidate predicates, you need to evaluate each one systematically. This is not a "pick the one that looks closest" exercise. It requires structured analysis across multiple dimensions.

Evaluation Criterion 1: Intended Use Alignment

This is the first and most important gate. The FDA's substantial equivalence analysis begins with intended use, and a mismatch here is fatal.

Same intended use means:

The same disease, condition, or purpose (diagnose, treat, prevent, monitor, mitigate)
The same patient population (pediatric vs. adult, acute care vs. home use)
The same body site or anatomical region (if applicable)
The same clinical context (prescription use, over-the-counter, professional use only)

Subset of intended use is also acceptable. If the predicate is cleared for a broader use and your device targets a narrower application within that scope, you can claim substantial equivalence. However, expanding beyond the predicate's intended use is the single fastest way to get an NSE determination.

Intended Use Comparison	Result
Your device and predicate have the same intended use	Proceed to technology comparison
Your intended use is a subset of the predicate's	Acceptable — proceed to technology comparison
Your intended use is broader than the predicate's	Problem — requires additional clinical data or a different predicate
Your intended use is fundamentally different	NSE — choose a different predicate or consider De Novo

Evaluation Criterion 2: Technological Characteristics

After confirming intended use alignment, evaluate the technological characteristics:

Materials and material composition — same biocompatible materials, or different?
Design and engineering specifications — similar form factor, dimensions, mechanical properties?
Energy source and type — same power source, same energy delivery mechanism?
Software and algorithms — similar functionality, or does your device introduce novel algorithms?
Manufacturing processes — any processes that could affect device performance differently?

Same technological characteristics as the predicate = straightforward SE argument. If your device uses the same materials, design approach, energy source, and software approach as the predicate, you likely need minimal additional performance data.

Different technological characteristics ≠ automatic NSE. The critical question is whether the differences raise new questions of safety or effectiveness:

If no — the device can still be found SE, but you need to explain why the differences are non-consequential
If yes — you must provide performance data (bench testing, animal studies, or clinical data) demonstrating that your device is as safe and effective as the predicate despite the differences

Evaluation Criterion 3: Regulatory History and Safety Profile

The FDA's draft "Best Practices for Selecting a Predicate Device" guidance recommends evaluating four factors:

Cleared using well-established methods — Was the predicate cleared based on testing methods and standards that are still current and accepted?
Meets or exceeds expected safety and performance — Does the predicate have a track record of safe and effective use?
Without unmitigated use-related or design-related safety issues — Check MAUDE for adverse event patterns. A predicate with a known safety signal that has not been resolved is a risky choice.
Without an associated design-related recall — A predicate recalled for a design defect raises questions about whether the predicate itself sets the right safety bar.

FDA's recommendation: The draft guidance suggests including a table in your 510(k) submission showing how each candidate predicate was evaluated against these four best practices, and explaining why you selected the predicate you did. This proactive documentation strengthens your submission.

Evaluation Criterion 4: Practical Considerations

Beyond the regulatory criteria, several practical factors influence predicate selection:

Predicate age:

Recent predicates (cleared within the last 5-10 years) are generally preferable. They reference current standards, provide more detailed public summaries, and face less reviewer skepticism.
The FDA initially considered requiring disclosure when a predicate is older than 10 years. They ultimately decided against an age-based rule, recognizing that older devices may have extensive long-term safety data. However, reviewers still view very old predicates with caution.
If your best predicate is decades old, be prepared to explain why no more recent alternative exists and why the comparison remains valid.

Availability of public information:

Predicates with public 510(k) Summaries give you detailed information about testing, standards, and SE rationale
Predicates with only a 510(k) Statement provide minimal public information, making it harder to plan your submission

Predicate chain length:

How many generations of predicates does this device trace through? Long chains invite scrutiny about predicate creep (discussed below)
Shorter chains with direct comparisons are stronger

Same product code:

A predicate in the same product code simplifies your comparison and aligns with reviewer expectations
Cross-product-code predicates are valid but require additional justification

Substantial Equivalence: What It Means and How to Demonstrate It

Substantial equivalence (SE) is the legal standard that the entire 510(k) pathway is built upon. It does not mean identical. It means the new device is at least as safe and effective as the predicate.

The FDA's SE Flowchart

The FDA follows a defined decision tree (described in the guidance document "The 510(k) Program: Evaluating Substantial Equivalence in Premarket Notifications [510(k)]"):

Decision Point 1: Does the new device have the same intended use as the predicate?

If no → NSE (unless clinical data demonstrates the different intended use does not raise new safety/effectiveness questions)
If yes → proceed to Decision Point 2

Decision Point 2: Does the new device have the same technological characteristics as the predicate?

If yes → SE (proceed to verification)
If no → proceed to Decision Point 3

Decision Point 3: Do the different technological characteristics raise new types of safety or effectiveness questions?

If no → SE (proceed to verification)
If yes → proceed to Decision Point 4

Decision Point 4: Can the submitter demonstrate through accepted scientific methods (bench testing, animal testing, clinical data) that the device is as safe and effective as the predicate?

If yes → SE
If no → NSE

Decision Point 5: Are the performance data acceptable?

If yes → SE determination issued
If no → NSE determination issued

Same vs. Different Intended Use

The FDA distinguishes between "intended use" and "indications for use":

Intended use is the general purpose of the device (e.g., "monitoring blood oxygen saturation")
Indications for use are the specific conditions or contexts for which the device is labeled (e.g., "continuous monitoring of arterial blood oxygen saturation in adult patients during surgery")

For SE purposes, the intended use must be the same. Indications for use can differ somewhat — for example, your device may target a subset of the predicate's indications — as long as the overall intended use is the same.

Expanding intended use beyond the predicate's scope is one of the most common grounds for NSE. If your device has a broader or different intended use, you have limited options:

Find a different predicate with a matching intended use
Provide clinical data supporting the different intended use
Pursue the De Novo pathway if no predicate exists for your intended use

Same vs. Different Technological Characteristics

"Same technological characteristics" means the device uses the same materials, design, energy source, and performance features as the predicate. In practice, very few new devices have truly identical technological characteristics — innovation inherently introduces differences.

The important question is not whether differences exist, but what kind of differences they are:

Type of Difference	Impact	What You Need
Same materials, minor dimensional changes	Low — generally does not raise new questions	Explanation of why the difference is non-consequential
Different material, same function	Moderate — may raise biocompatibility questions	Biocompatibility testing per ISO 10993
Different energy source	Moderate to high — depends on application	Performance and safety testing specific to the new energy type
Novel software algorithm	Moderate to high — depends on clinical impact	Software validation per IEC 62304, possibly clinical performance data
Fundamentally different mechanism of action	High — likely raises new questions	Comprehensive performance data, possibly clinical studies; consider De Novo

When Different Technology Requires Performance Data

If your technological characteristics differ from the predicate and those differences raise new questions of safety or effectiveness, the FDA requires performance data to resolve those questions. The type and extent of data depend on the nature of the differences:

Bench testing: The most common requirement. Mechanical testing, electrical safety testing, biocompatibility testing, software testing, sterilization validation — whatever is needed to demonstrate that your device performs as safely and effectively as the predicate despite the technological differences.

Animal testing: Required in some cases, particularly for implantable devices or devices that deliver energy to tissue. Animal studies can demonstrate in vivo performance and biocompatibility when bench testing alone is insufficient.

Clinical data: Required when the technological differences are significant enough that bench and animal data alone cannot resolve safety and effectiveness questions. The FDA's guidance on clinical data in 510(k) submissions identifies four scenarios where clinical data may be necessary:

When the new device's indications for use differ from the predicate
When the technological characteristics of the new device vary significantly from the predicate
When substantial equivalence cannot be established through non-clinical testing alone
When newly discovered or escalated risks associated with the predicate device indicate a need for clinical data

Specific scenarios requiring clinical data:

Different patient population: If your device targets a pediatric population but the predicate was cleared for adults only, clinical data demonstrating safety and performance in the pediatric population is typically required. Similarly, expanding from professional-use to home-use contexts may require clinical usability data.
Different materials in body-contact devices: A predicate uses non-absorbable surgical material, but your device uses a bioabsorbable material. The body's absorption of the material raises new biological questions that cannot be fully resolved through bench testing alone — clinical or animal data demonstrating in vivo performance is typically needed.
No non-clinical testing model available: When no validated bench testing model or phantom exists to simulate the clinical environment (e.g., a specific anatomical site for an imaging device), real clinical data may be the only way to demonstrate performance equivalence.
Post-market safety signals in the predicate: If new adverse event data has emerged for the predicate device since its clearance — particularly signals that were not considered in the original review — the FDA may require clinical data from the new device to address these newly identified risks.

Real-World Substantial Equivalence Scenarios

Scenario A — Straightforward SE (same intended use, same technology): A company develops a new digital sphygmomanometer (blood pressure monitor) with the same oscillometric measurement technology, same intended use (non-invasive blood pressure measurement for home use), and same materials as a recently cleared predicate. The only difference is a smaller form factor and updated display. This is a textbook SE determination — same intended use, same technological characteristics, no new questions of safety or effectiveness.

Scenario B — Different technology, same intended use, no new safety questions: A company develops a new pulse oximeter using a different LED wavelength combination than the predicate. The intended use is identical (continuous SpO2 monitoring during surgery). The different LED wavelengths are a technological difference, but they do not raise new questions of safety or effectiveness — the measurement principle is the same, and the company provides bench testing showing equivalent accuracy. SE determination is appropriate.

Scenario C — Different technology, same intended use, new safety questions requiring data: A company develops a cardiac monitor that uses a novel flexible electrode material instead of the predicate's traditional Ag/AgCl electrodes. The intended use is identical, but the new material raises biocompatibility questions (skin contact over extended periods) and signal quality questions. The company must provide biocompatibility testing per ISO 10993 and signal quality testing demonstrating equivalent performance. With adequate data, SE is achievable.

Scenario D — Different intended use, no viable predicate: A company develops a wearable device that monitors hydration levels through skin impedance — a use for which no cleared device exists. The closest available devices are skin impedance measurement devices cleared for body composition analysis, but this is a different intended use. The 510(k) pathway is not appropriate. The company should pursue De Novo or seek FDA guidance through a Pre-Submission meeting.

When Predicates Are NOT Equivalent — Illustrative Counter-Examples

Understanding why certain device pairs fail substantial equivalence is just as important as understanding when they succeed. These examples illustrate common misjudgments:

Forehead thermometer vs. ear (tympanic) thermometer: Despite both measuring body temperature, a forehead thermometer and an ear thermometer are generally not appropriate predicates for each other. The ear thermometer involves insertion into the ear canal, raising biocompatibility concerns (contact with mucosal tissue) and usability risks (potential injury to the tympanic membrane) that are entirely absent with a forehead measurement device. The different body site creates new safety questions that prevent a straightforward SE determination.

External airway management device vs. oropharyngeal airway: An external neck device that applies continuous vacuum pressure to manage the airway is not substantially equivalent to an oropharyngeal tube — even though both devices serve the broader function of airway management. The external device applies sustained pressure to soft tissue and nerves, raising new risks of nerve damage and tissue injury that do not exist with an inserted oropharyngeal device. The fundamentally different mechanism of action creates new safety questions.

AI software analyzing different anatomical regions: An AI-powered image analysis tool cleared for detecting anomalies in breast MR imaging may or may not be a valid predicate for a similar AI tool targeting a different anatomical region (e.g., liver, brain). Even when the underlying AI technology is the same, expanding to a different anatomical region may constitute a different intended use — particularly when the clinical significance, patient population, or risk profile differs. The FDA has accepted broad interpretations of "same intended use" for AI/ML devices in some cases (e.g., "assessment of specific disease states using machine learning"), but this requires careful justification and often clinical data for the new anatomical region.

The lesson: Similarity in broad function does not equal substantial equivalence. The SE determination depends on specific intended use, body site, patient contact, and whether the differences raise new questions of safety or effectiveness. When in doubt, a Pre-Submission meeting can clarify whether the FDA considers two devices comparable.

Using Multiple Predicates: Strategy and Documentation

While every 510(k) must have one clearly identified primary predicate, using additional predicates strategically can strengthen your submission. The key is doing it correctly.

When Multiple Predicates Make Sense

Your device combines features found in different cleared devices — as long as all cited predicates share the same intended use
You want to demonstrate regulatory precedent for a specific material, technology, or feature that your primary predicate does not incorporate
Your device has a broader feature set than any single predicate — reference predicates can show that each individual feature has been previously cleared
Your device is a multi-parameter system — for example, a patient vital signs monitor that measures temperature, blood pressure, and heart rate may cite separate predicates for each measurement parameter, as long as all predicates share the same overall intended use (patient monitoring). The primary predicate should be the device most similar to your system overall, with additional reference predicates supporting individual parameters.

How to Document Multiple Predicates

In your submission, be explicit about the role each predicate plays:

Clearly designate the primary predicate — state the 510(k) number, device name, manufacturer, and intended use
Explain the purpose of each reference predicate — state specifically what aspect of your SE argument each reference predicate supports
Demonstrate that all predicates share the same intended use — this is the line the FDA draws between acceptable multi-predicate submissions and prohibited split predicates
Provide a comparison table for each predicate — show the specific features or characteristics being compared

Example documentation structure:

Predicate	Role	510(k) Number	What It Supports
Acme Pulse Ox Pro	Primary	K231234	Overall SE — same intended use, same core technology
Beta Medical OxSensor	Reference	K201567	Demonstrates prior clearance of novel optical sensor material
Gamma Health ConnectOx	Reference	K221890	Demonstrates prior clearance of BLE wireless connectivity in monitoring devices

Predicate Selection and 510(k) Submission Types

Your predicate selection is intertwined with which type of 510(k) submission you file. The FDA recognizes three types, each with different predicate implications.

Traditional 510(k)

The Traditional 510(k) is the standard submission type. You identify a predicate device, build a full substantial equivalence argument, and include all comparison data, performance testing, and documentation. Most original 510(k) submissions are Traditional.

Use a Traditional 510(k) when:

You are submitting a new device for the first time
Your device has technological differences from the predicate that require performance data
You are referencing a competitor's device as your predicate

Special 510(k)

A Special 510(k) is available only when a manufacturer is making modifications to its own legally marketed device. The predicate is your own previously cleared device. This pathway is faster (typically reviewed within 30 days) because the FDA already has your prior submission on file.

Use a Special 510(k) when:

You are modifying your own previously cleared device
The modification does not change the device's intended use
The modification does not alter the fundamental scientific technology of the device
You can rely on design controls, risk analysis, and verification/validation testing to demonstrate that the modification does not adversely affect safety or effectiveness

Connection to predicate selection: The Special 510(k) is the practical reason many companies keep their own prior clearances as predicates. If your modification qualifies for a Special 510(k), you can leverage the streamlined review. If your changes are too significant for the Special pathway, you will need to file a Traditional 510(k) — and you should evaluate whether your own prior device or a competitor's more recently cleared device provides the stronger SE argument.

Abbreviated 510(k)

An Abbreviated 510(k) relies on FDA guidance documents, special controls, or recognized consensus standards to establish substantial equivalence. Instead of providing full test reports for every performance characteristic, you demonstrate compliance with the applicable guidance or standard and reference the predicate.

Use an Abbreviated 510(k) when:

An FDA-recognized guidance document or special controls document exists for your device type
Your device complies with the relevant recognized consensus standards
The guidance or standards adequately address the differences between your device and the predicate

Practical note: The Abbreviated 510(k) does not change which predicate you select — it changes how you demonstrate substantial equivalence. Your predicate evaluation criteria remain the same. The advantage is reduced documentation burden when well-established guidance and standards exist for your device category.

Using De Novo-Authorized Devices as Predicates

Devices that receive marketing authorization through the De Novo classification process under Section 513(f)(2) of the FD&C Act can serve as predicates for subsequent 510(k) submissions — provided they are not exempt from premarket notification requirements.

How It Works

When a novel device is authorized through De Novo, the FDA creates a new classification regulation (a new product code and regulatory classification). This newly classified device then becomes the first predicate in that classification, enabling future manufacturers of similar devices to use the 510(k) pathway rather than filing their own De Novo request.

Key Considerations

Verify that the De Novo device is not exempt from 510(k): Some De Novo classifications result in exemption from premarket notification. If the device type is exempt, it cannot serve as a predicate.
Review the classification order: The De Novo authorization includes a classification order that specifies the device type, product code, and special controls. Your device must fit within this classification to use the De Novo device as a predicate.
Expect evolving requirements: De Novo-authorized devices often establish new special controls. The FDA may expect your 510(k) submission to address these special controls even if the predicate's own De Novo submission had different requirements.

Practical Implications

De Novo-authorized predicates are becoming increasingly common, particularly in digital health, AI/ML software devices, and novel diagnostic technologies. If you are entering a relatively new device category, check whether a De Novo authorization exists — it may provide a 510(k) pathway where one did not exist before.

Example: A company develops an AI-powered dermatology screening tool. No traditional 510(k) predicate exists. However, a competitor recently received De Novo authorization for a similar AI dermatology device (DEN210xyz), creating product code "QXZ" with specified special controls. The new company can now file a 510(k) referencing the De Novo device as a predicate — a significantly faster and less expensive path than filing their own De Novo.

Using Rescinded or Withdrawn 510(k) Devices as Predicates

This is one of the most frequently misunderstood areas of predicate selection. The short answer: it depends on why the device was withdrawn.

Devices You CAN Use as Predicates

A device can still serve as a predicate if:

The 510(k) clearance was voluntarily rescinded by the manufacturer (e.g., the company stopped marketing the device for business reasons)
The device was withdrawn from the market for commercial reasons (not safety or effectiveness concerns)
The device was recalled — even a Class I recall does not automatically invalidate the device as a predicate

The key legal question is whether the device was legally marketed at some point. If it was, and it was not removed by FDA enforcement action or found misbranded/adulterated by court order, it can technically serve as a predicate.

Devices You CANNOT Use as Predicates

A device cannot serve as a predicate if:

It was removed from the market at FDA's initiative (e.g., through an FDA-initiated seizure or injunction)
It was determined to be misbranded or adulterated by judicial order
It was never legally marketed in the first place

Practical Guidance

Even when a rescinded or withdrawn device is technically eligible as a predicate, think carefully before using one:

Reviewers will ask questions. If a device is no longer on the market, reviewers may wonder why — and may question whether the comparison is appropriate.
Limited public information. A device withdrawn years ago may have minimal available documentation, making your comparison harder.
The FDA's "Best Practices" draft guidance discourages it. While not yet finalized, the guidance encourages selecting predicates that are currently marketed, have a strong safety record, and use current methods and standards.
A better predicate likely exists. In most cases, a currently marketed device with a recent clearance will serve your submission better.

When it makes sense: In rare cases, a rescinded 510(k) may be the best or only predicate available — for example, in a niche device category with very few cleared devices. If this is your situation, document your predicate search thoroughly and consider a Pre-Submission meeting to confirm FDA agreement.

Predicate Chains and Predicate Creep

What Is Predicate Creep?

Predicate creep (also called "predicate drift") occurs when successive generations of 510(k) clearances gradually drift from the original predicate through accumulated small changes in technological characteristics — each generation slightly different from the last, until the current device bears little resemblance to the original predicate in the chain.

A 2023 study published in PLOS ONE examined predicate creep in robotic surgical devices and found that devices cleared through long predicate chains could trace their lineage back to manually operated instruments with fundamentally different technology, raising questions about whether the substantial equivalence standard was truly met across the full chain.

Why FDA Is Concerned

The FDA has identified predicate creep as a systemic concern with the 510(k) program. The problem is not any individual clearance in the chain — each step may have been a reasonable SE determination at the time. The problem is the cumulative effect: a device cleared in 2025 may be "substantially equivalent" to a 2020 device, which was "substantially equivalent" to a 2015 device, which traces back to a 1990 device, which traces back to a 1976 pre-amendment device. The 2025 device and the 1976 device may share almost nothing in common.

How to Avoid Predicate Creep Issues

Choose the most direct comparison available. If a recently cleared device with the same intended use and similar technology exists, use it — do not chain through intermediary devices.
Minimize the number of "hops" in your predicate chain. Each generation of difference adds scrutiny.
Be transparent about the chain. If your predicate itself has a long predicate history, acknowledge it. Reviewers can look up the chain themselves.
Justify accumulated differences. If your device has drifted technologically from the original device in the chain, be prepared to explain why the current comparison (to your immediate predicate) remains valid.

Practical reality: The FDA evaluates your device against your immediate predicate — not against every device in the chain. But if a reviewer notices significant drift, they may ask additional questions. The best defense is choosing a predicate that minimizes the need for chain-tracing.

What to Do When No Predicate Exists

If your systematic search of FDA databases reveals no legally marketed device with the same intended use and comparable technological characteristics, you do not have a valid predicate. Attempting to force-fit a 510(k) with an inappropriate predicate will result in an NSE determination or an RTA (Refuse to Accept) hold.

Automatic Class III Classification

Under Section 513(f)(1) of the FD&C Act, a new device type that does not have a valid predicate is automatically classified as Class III — regardless of its actual risk level. This is a statutory default: if the device is not within a type that was on the market before the 1976 Medical Device Amendments or that has been classified into Class I or II, it is classified into Class III by operation of law. This automatic Class III classification means the device requires Premarket Approval (PMA) unless the manufacturer pursues a De Novo request to reclassify it into Class I or II.

This is a critical point for novel device developers: the absence of a predicate does not just mean you cannot use the 510(k) pathway — it means your device faces the highest regulatory bar by default. Understanding this automatic classification mechanism is essential when deciding whether to invest in a predicate search, pursue De Novo, or plan for PMA.

Your Options

Option 1: De Novo Classification Request The De Novo pathway exists specifically for novel, low-to-moderate-risk devices that lack a predicate. A successful De Novo creates a new classification regulation and product code, and your device becomes the first predicate for that device type.

Advantages: Creates a pathway for your device and for future competitors (which may strengthen your market position as the "first" in the category)
Disadvantages: Longer review timeline (typically 150-250 days), higher user fees, and a more rigorous standard of review (reasonable assurance of safety and effectiveness, rather than substantial equivalence)

Option 2: PMA (Premarket Approval) If your device is high-risk (Class III), and no predicate exists, the PMA pathway may be required. PMA requires an independent demonstration of reasonable assurance of safety and effectiveness — typically involving clinical trials — and is significantly more expensive and time-consuming than either 510(k) or De Novo.

Option 3: Re-evaluate your intended use In some cases, narrowing or adjusting your intended use statement may bring your device within the scope of an existing predicate's clearance. This is a legitimate strategy — but only if your actual intended use genuinely fits within the predicate's scope. Do not misrepresent your intended use to fit a predicate.

Option 4: Pre-Submission (Q-Sub) meeting If you are uncertain whether a predicate exists or which pathway is appropriate, a Pre-Submission meeting with the FDA can provide clarity. The FDA will review your device description, proposed intended use, and any candidate predicates, and provide written feedback on the appropriate regulatory pathway.

Do not: Try to "create" a predicate by arguing that a device from a fundamentally different category is analogous to yours. Reviewers see through this immediately, and it wastes months of everyone's time.

If You Receive an NSE Determination

If the FDA issues a Not Substantially Equivalent (NSE) determination, you have several formal options:

Resubmit a new 510(k) with a different predicate — if you identify a more appropriate predicate, you can file a new submission with a revised SE argument and additional supporting data
Resubmit with additional data — if the predicate itself was reasonable but your original submission lacked sufficient evidence, you can resubmit the same predicate with additional performance, biocompatibility, or clinical data to address the FDA's concerns
File a De Novo classification request — under Section 513(f)(2), you can submit a De Novo request directly, without needing a prior NSE determination (since FDASIA modified the process in 2012). If successful, the FDA reclassifies your device into Class I or II and creates a new product code
File a reclassification petition — under Section 513(e), you can petition the FDA to reclassify the device type from Class III to Class I or II based on available safety and effectiveness information
Submit a PMA — if your device genuinely requires Class III oversight, proceed with a Premarket Approval application

Timing matters: An NSE determination does not prevent you from taking any of these actions, but each one requires significant additional time and investment. This is why getting predicate selection right — or confirming it through a Pre-Submission meeting — is so valuable upfront.

Pre-Submission (Q-Sub) for Predicate Agreement

A Pre-Submission (also called a Q-Submission or Q-Sub) is one of the most valuable tools available for predicate device selection. It is free, and the FDA provides written feedback.

When to Use a Pre-Sub for Predicate Issues

You are unsure which predicate to select and want FDA input
Your best candidate predicate has a different product code and you want to confirm it is acceptable
Your device has significant technological differences from the predicate and you want to confirm the testing approach
You are considering multiple predicates and want FDA agreement on the predicate strategy
You want to confirm whether your device is appropriate for the 510(k) pathway at all (vs. De Novo or PMA)

What to Include in Your Pre-Sub Package

Device description — detailed description of your device, its intended use, and how it works
Proposed intended use / indications for use statement
Proposed predicate device(s) — 510(k) number, device name, manufacturer, intended use, and product code for each
Preliminary comparison table — side-by-side comparison of your device vs. the proposed predicate, highlighting similarities and differences
Specific questions — focused, answerable questions such as:
- "Does the FDA agree that [K######] is an appropriate predicate for this device?"
- "Does the FDA agree that the intended use of our device is the same as the predicate's?"
- "Given the technological differences described above, what testing does the FDA recommend?"

Pre-Sub Timeline and Process

Submit the Pre-Sub package through the Document Control Center (DCC)
FDA screens the submission within approximately 15 days
FDA provides written feedback within approximately 70-75 days
Meeting (if requested): Teleconference, in-person, or written-only, typically scheduled around day 70-75

Why Pre-Subs Are Worth the Wait

The 70-75 day timeline may seem long, but consider the alternative: submitting a 510(k) with an uncertain predicate selection, receiving an AI request at day 60 asking you to justify your predicate choice, and potentially facing an NSE determination months later. A Pre-Sub invests 2-3 months upfront to avoid 6-12 months of rework.

Key benefit: FDA feedback from a Pre-Sub is documented in writing. If you follow the agreed-upon predicate and testing strategy, reviewers are unlikely to deviate from it during the 510(k) review. This provides a high degree of predictability.

Documenting Your Predicate Selection Rationale

Strong documentation of your predicate selection rationale serves two purposes: it demonstrates to the FDA that you conducted a thorough, systematic search; and it protects your internal decision-making if questions arise later.

What to Include in Your Submission

1. Summary of predicate search methodology

Which databases you searched (510(k) database, Product Classification database, GUDID, MAUDE)
What search terms and filters you used
The date range of your search
How many candidate devices you identified

2. Candidate predicate evaluation matrix

List all candidate devices considered
For each, document intended use, technology, product code, clearance date, and reason for selection or rejection

3. Primary predicate justification

Why this predicate is the best match for your device
How the intended use aligns
How the technological characteristics compare
Any differences and why they do not raise new safety or effectiveness questions

4. Reference predicate justification (if applicable)

The specific purpose each reference predicate serves
Confirmation that each reference predicate shares the same intended use

5. Predicate selection narrative in the 510(k) Summary

The FDA's draft "Best Practices" guidance recommends that submitters include a narrative within the 510(k) Summary explaining their predicate selection rationale. This narrative should describe how you applied the four best practices (well-established methods, safety and performance record, no unmitigated safety issues, no design-related recalls) to arrive at your selected predicate. Including this narrative proactively demonstrates thoroughness and may reduce the likelihood of AI requests related to predicate selection.

The Predicate Comparison Table: What FDA Expects

The comparison table is one of the most scrutinized elements of a 510(k) submission. The FDA expects a structured, row-by-row comparison covering at minimum:

Characteristic	Subject Device	Predicate Device (K######)	Comparison / Discussion
510(k) number	N/A (new submission)	K######	—
Manufacturer	[Your company]	[Predicate manufacturer]	—
Trade name	[Your device name]	[Predicate trade name]	—
Product code	[Code]	[Code]	Same / Different (with explanation)
Intended use / Indications for use	[Your IFU statement]	[Predicate's cleared IFU]	Same / Subset — explain
Device description	[Detailed description]	[From predicate summary]	Similar — differences noted below
Materials	[Specific materials]	[Predicate materials]	Same / Different (with justification)
Energy source	[Type]	[Type]	Same / Different
Software / algorithms	[Description]	[Description]	Same / Different
Dimensions / size range	[Specifics]	[Specifics]	Within range / Different
Sterility	[Method]	[Method]	Same / Different
Biocompatibility	[Testing performed]	[Testing referenced]	Equivalent
Performance specifications	[Your data]	[Predicate data]	Meets or exceeds
Labeling	[Included]	[Referenced]	Consistent

Formatting tip: Do not leave any cells blank. If information is not available for the predicate, state "Not available in public summary" rather than leaving the cell empty. Empty cells invite questions.

The 7-Step Predicate Selection Workflow

Here is a consolidated, practical workflow for selecting a predicate device from initial search to final documentation.

Step 1: Define Your Device Profile

Before you search, clearly define:

Your device's intended use and indications for use
The technology, materials, design, and energy source
The clinical context (who uses it, where, on whom)
Your target product code(s)

This profile becomes your search framework. Without it, you will waste time evaluating irrelevant candidates.

Step 2: Conduct a Comprehensive Database Search

Search the FDA's databases systematically:

Product Classification Database → confirm product code and classification
510(k) Database → search by product code, applicant, device name
GUDID → cross-reference commercial names with 510(k) numbers
MAUDE → check adverse event history for candidates
Recall Database → check for recall history

Cast a wide net initially. You can narrow down later. Document every search — dates, terms, filters, results.

Step 3: Build a Candidate Shortlist

From your search results, compile a shortlist of 5-10 candidate predicates. For each, record:

510(k) number, device name, manufacturer
Product code and classification
Decision date
Intended use (from the 510(k) summary or clearance letter)
Key technological characteristics (if available from the summary)
Whether a public 510(k) summary exists

Step 4: Evaluate Each Candidate Against Selection Criteria

Apply the evaluation criteria systematically:

Intended use match — same, subset, or different?
Technological characteristics — same, similar with minor differences, or fundamentally different?
Safety profile — any adverse events, recalls, or enforcement actions?
Practical factors — age of clearance, availability of public summary, predicate chain length, same product code?

Eliminate candidates that fail the intended use gate. Rank remaining candidates by overall fit.

Step 5: Select Your Primary Predicate (and Reference Predicates, If Needed)

Choose the candidate that provides the strongest overall SE argument:

Closest intended use match
Fewest technological differences requiring justification
Strong safety record
Recent clearance with detailed public summary
Same product code (preferred, not required)

If no single predicate covers all aspects of your device, identify reference predicates to support specific features — ensuring all predicates share the same intended use.

Step 6: Validate with a Pre-Submission (If Warranted)

File a Pre-Sub if:

You have any doubt about whether your predicate is acceptable
Your device has significant technological differences from the predicate
You are using a predicate from a different product code
You are entering a new device category with limited predicate history

The 70-75 day investment for a Pre-Sub is almost always worthwhile for first-time submitters or novel devices.

Step 7: Document Everything

Prepare your predicate selection rationale documentation:

Search methodology summary
Candidate evaluation matrix
Primary predicate justification
Reference predicate justification (if applicable)
Comparison table

This documentation feeds directly into your 510(k) submission's substantial equivalence section.

Common Predicate Selection Mistakes

Mistake 1: Choosing a Predicate with a Different Intended Use

This is the most fundamental error. If your device monitors blood glucose for home use and your predicate monitors blood glucose for clinical laboratory use, the intended uses are different — different clinical context, different user population, different risk profile. The FDA will identify this immediately.

How to avoid it: Read the predicate's cleared indications for use statement word by word. Compare it to your own intended use. Any material difference in disease/condition, patient population, body site, or clinical context is a red flag.

Mistake 2: Ignoring the Predicate's Safety History

Selecting a predicate without checking MAUDE and the recall database is like buying a house without an inspection. If the predicate has a pattern of adverse events or a design-related recall, reviewers will notice — and they may question whether the predicate represents an appropriate safety baseline.

How to avoid it: Always check MAUDE and the recall database before committing to a predicate. If the predicate has known safety issues, either choose a different predicate or be prepared to proactively address those issues in your submission.

Mistake 3: Using a Very Old Predicate Without Justification

A predicate from 1985 may be legally valid, but it invites skepticism. Standards have changed. Testing expectations have evolved. Review rigor has increased. An old predicate may not represent the current state of the art, and reviewers may question why no more recent predicate was available.

How to avoid it: Prefer predicates cleared within the last 5-10 years. If you must use an older predicate, document why no more recent alternative exists and why the comparison remains valid. Consider supplementing with reference predicates that are more recent.

Mistake 4: Relying on Split Predicates

As discussed above, split predicates — combining intended use from one device with technological characteristics from a device with a different intended use — are prohibited. Some submitters still attempt this, particularly for devices that combine features from different clinical domains.

How to avoid it: Ensure every predicate you cite shares the same intended use as your device. If no single predicate or combination of same-intended-use predicates covers your device's functionality, the 510(k) pathway may not be appropriate. Consider De Novo.

Mistake 5: Failing to Document the Predicate Search

Submitting a 510(k) without demonstrating that you conducted a thorough predicate search is a missed opportunity. The FDA's draft "Best Practices" guidance encourages submitters to document their search methodology and explain why the selected predicate is the best available option.

How to avoid it: Keep detailed records of your database searches from the beginning. Include search methodology and a candidate evaluation matrix in your submission.

Mistake 6: Not Using a Pre-Submission When the Predicate Is Uncertain

If you have any doubt about whether your predicate is acceptable, a Pre-Submission meeting is the appropriate tool. Submitting a 510(k) with an uncertain predicate and hoping for the best is a recipe for AI requests and potential NSE.

How to avoid it: File a Pre-Sub. It is free. The FDA will tell you in writing whether your predicate is appropriate. This is the single most cost-effective risk mitigation tool available.

Mistake 7: Citing Too Many Predicates Without Clear Purpose

Some submitters attempt to strengthen their argument by citing five, six, or more predicates. This often backfires. Each additional predicate adds complexity to the SE argument, and if the reviewer cannot quickly understand why each predicate is cited, it creates confusion — or worse, the impression that no single predicate truly matches the device.

How to avoid it: Limit your predicates to those that serve a clear, specific purpose. One primary predicate and one or two reference predicates is usually sufficient. For each predicate cited, you should be able to articulate in one sentence why it is included.

Mistake 8: Choosing a Predicate Solely Because It Is Your Own Device

If you hold a previous 510(k) clearance, it is natural to consider your own prior device as the predicate. This is often a good strategy — but not always. If your modification is substantial and changes the device's technological characteristics significantly, a competitor's more recently cleared device with closer technology might actually be a better primary predicate.

How to avoid it: Evaluate all candidates objectively, including your own prior devices and competitors' devices. The best predicate is the one that provides the strongest SE argument, regardless of manufacturer.

Predicate Selection Considerations by Device Type

Different device categories present distinct predicate selection challenges. Here are considerations for several common device types.

Software as a Medical Device (SaMD) and AI/ML Devices

The number of AI/ML-enabled devices cleared through 510(k) has grown rapidly, with over 1,000 AI/ML devices authorized by the FDA as of 2024
Many AI/ML devices reference other AI/ML devices as predicates, even when the underlying algorithms differ significantly
Focus on matching the clinical function (e.g., computer-aided detection, triage, measurement) rather than the specific AI/ML architecture
Software-specific guidance documents (IEC 62304, FDA's software guidance, cybersecurity guidance) add testing requirements regardless of predicate selection
For SaMD, see our SaMD Regulatory Guide and IEC 62304 Guide

Implantable Devices

Biocompatibility matching is critical — even minor material differences from the predicate require ISO 10993 testing
Implant duration (transient, short-term, long-term/permanent) must match the predicate's
Body contact category (surface, externally communicating, implant) must align
Mechanical performance standards (ASTM, ISO) are well-established for many implant categories, making comparison straightforward
Long-term safety data for the predicate can strengthen your submission

Combination Products (Device + Drug or Device + Biologic)

If the device component has a valid predicate, a 510(k) may be appropriate for the device component — but the combination product as a whole involves additional regulatory complexity
The lead center (CDRH, CDER, or CBER) determines the primary pathway
A Pre-Submission meeting is strongly recommended for any combination product to clarify predicate applicability and testing expectations

In Vitro Diagnostic (IVD) Devices

Predicate selection for IVDs focuses heavily on the analyte measured, the specimen type, and the clinical purpose (screening, diagnosis, monitoring)
Analytical performance characteristics (sensitivity, specificity, accuracy, precision) are key comparison points
Reference to recognized consensus standards (e.g., CLSI guidelines) can support the SE argument
For IVDs with novel biomarkers, the 510(k) pathway may not be available — De Novo is more common

FDA's Perspective on Predicate Selection

Key FDA Guidance Documents

The following FDA guidance documents directly address predicate device selection. Reading these documents is essential for anyone preparing a 510(k) submission.

Guidance Document	Key Content
The 510(k) Program: Evaluating Substantial Equivalence in Premarket Notifications [510(k)] (2014, final)	The foundational guidance — defines the SE flowchart, decision points, and evaluation criteria
Best Practices for Selecting a Predicate Device to Support a Premarket Notification [510(k)] Submission (2023, draft)	Recommends four best practices for predicate selection; proposes documentation expectations
How to Find and Effectively Use Predicate Devices (FDA webpage)	Practical instructions for searching FDA databases
Refuse to Accept Policy for 510(k)s (updated)	Defines the minimum acceptance criteria including predicate-related requirements
Requests for Feedback and Meetings for Medical Device Submissions: The Q-Submission Program (2023, final)	Guidance on Pre-Submission process for predicate agreement

What Reviewers Look For

Based on published FDA guidance and publicly available review practices, FDA reviewers evaluate predicate selection by asking:

Is the predicate legally marketed?
Does the predicate have the same intended use as the subject device?
Are the technological characteristics the same or different?
If different, do the differences raise new questions of safety or effectiveness?
Has the submitter addressed any known safety issues with the predicate?
Is the predicate comparison table complete and accurate?
Is the indications for use statement consistent across the submission (cover letter, Form FDA 3881, device description, comparison table, labeling)?

Inconsistency in the intended use statement across different sections of the submission is one of the most common RTA triggers. Ensure that your intended use statement is identical everywhere it appears.

About Form FDA 3881: The Indications for Use Statement (Form FDA 3881) is the formal FDA form that captures your device's indications for use and specifies whether the device is intended for prescription use or over-the-counter (OTC) use. This form must be included in every 510(k) submission and must be consistent with all other intended use references throughout your filing. Many RTA holds stem from discrepancies between the Form 3881 statement and the intended use described in other sections.

The Least Burdensome Approach

The FDA applies a "least burdensome" principle to 510(k) reviews, meaning that the amount and type of information required to demonstrate substantial equivalence should be the minimum necessary to address the regulatory question. If recognized consensus standards, FDA guidance documents, or established test methods adequately address the differences between your device and the predicate, the FDA will not require additional or duplicative testing beyond what these resources cover.

In practice, this means you should:

Reference specific recognized consensus standards (ISO, IEC, ASTM) your device complies with
Cite applicable FDA guidance documents that address your device type
Provide summary-level information when a full test report is not necessary to resolve the SE question
Include full test reports only where the differences from the predicate raise questions not addressed by existing standards or guidance

Understanding the least burdensome principle can help you scope your testing program appropriately — do not over-test, but do not under-test either.

FDA Substantial Equivalence vs. EU MDR Equivalence

If you are pursuing both US and EU market access, understanding the differences between FDA substantial equivalence and EU MDR clinical equivalence is important. While both concepts involve comparing your device to an existing device, they differ significantly in scope, criteria, and strictness.

Aspect	FDA 510(k) Substantial Equivalence	EU MDR Clinical Equivalence
Scope	Entire conformity assessment — the predicate comparison is the basis for market clearance	Clinical evaluation only — equivalence is one component of the broader conformity assessment process
Evaluation criteria	Intended use + technological characteristics	Technical characteristics + biological characteristics + clinical characteristics (all three must be demonstrated)
Material requirements	Flexibility allowed — different materials are acceptable if they do not raise new safety/effectiveness questions	Stricter — materials must be essentially identical, particularly for implantable and invasive devices
Access to comparator data	Public FDA databases provide substantial information about cleared devices	Manufacturers typically need access to the comparator's technical documentation or a contract with the comparator manufacturer, especially for Class III and implantable devices
Regulatory standard	Substantial equivalence (as safe and effective as the predicate)	Sufficient clinical evidence to demonstrate safety and performance, with equivalence used to leverage existing clinical data
Overall strictness	More permissive — allows broader technology differences with supporting data	Stricter per MEDDEV 2.7/1 rev.4 and MDCG guidance — harder to claim equivalence for devices with material or design differences

Practical implication: A device that successfully demonstrates substantial equivalence for FDA 510(k) clearance may not be able to claim clinical equivalence under EU MDR. The EU's stricter requirements for biological and material equivalence, combined with the need for access to comparator device technical documentation, mean that your EU regulatory strategy may require independent clinical data even when your FDA strategy relies entirely on predicate comparison.

Predicate Selection Checklist

Use this checklist before finalizing your predicate selection:

Predicate Validity

The predicate is a legally marketed device (510(k)-cleared, De Novo-authorized, or pre-amendment)
The predicate was not removed from the market by FDA enforcement action
The predicate was not found misbranded or adulterated by judicial order
The predicate is not a PMA-approved device

Intended Use

The predicate's intended use is the same as your device's (or your device's is a subset)
The patient population matches (age group, clinical setting)
The body site / anatomical region matches (if applicable)
The clinical context matches (home use vs. professional, prescription vs. OTC)
Your indications for use statement is consistent with the predicate's scope

Technological Characteristics

You have identified all technological differences between your device and the predicate
For each difference, you have determined whether it raises new safety or effectiveness questions
For differences that raise new questions, you have a testing plan to generate the required performance data
You have documented why any differences are non-consequential (where applicable)

Safety and Regulatory History

You have searched MAUDE for adverse events associated with the predicate
You have searched the recall database for recalls associated with the predicate
The predicate does not have unmitigated design-related safety issues
The predicate does not have a design-related recall (or you have justified its use despite one)

Documentation

You have documented your search methodology (databases searched, terms used, dates)
You have a candidate evaluation matrix showing all predicates considered
You have a written justification for selecting the primary predicate
You have a completed comparison table with no blank cells
Your intended use statement is consistent across all submission sections

Pre-Submission

You have considered whether a Pre-Sub is warranted (recommended for uncertain predicate selections, novel devices, or first-time submitters)
If a Pre-Sub was conducted, FDA feedback supports your predicate selection

Frequently Asked Questions

Can I use a competitor's device as a predicate?

Yes. Any legally marketed device can serve as a predicate, regardless of who manufactures it. You do not need permission from the predicate holder. In fact, most 510(k) submissions reference competitors' devices as predicates. The FDA's 510(k) database is public, and 510(k) Summaries are public records.

Can I use a recalled device as a predicate?

Yes, but proceed with caution. A recalled device was still legally marketed at some point and technically remains a valid predicate. However, the FDA's draft "Best Practices" guidance recommends against selecting predicates with design-related recalls. If you use a recalled device, explain why it remains an appropriate comparator and how your device addresses the issue that led to the recall.

Does predicate age matter?

Legally, no — there is no age limit on predicates. Practically, yes — older predicates raise reviewer questions. The FDA considered a 10-year disclosure requirement but ultimately chose to recommend best practices rather than impose an age-based rule. Prefer recent predicates. If you must use an old one, justify the choice.

Can I use a device from a different product code as a predicate?

Yes, but you need to justify why the cross-product-code comparison is appropriate. The FDA evaluates SE based on intended use and technological characteristics, not product code. However, different product codes may indicate different intended uses or device types, so the justification bar is higher.

Can I change my predicate after submitting my 510(k)?

Changing the predicate after submission is essentially a new submission. The entire SE argument, comparison table, and testing rationale are built around the predicate. If the FDA determines your predicate is inappropriate during review, you will likely need to withdraw and resubmit with a different predicate — or file a new 510(k).

What if FDA disagrees with my predicate selection during review?

If the FDA questions your predicate during substantive review, they will typically issue an Additional Information (AI) request asking you to justify the selection or identify an alternative predicate. If you cannot satisfy the reviewer, the outcome may be an NSE determination. This is why a Pre-Submission meeting is so valuable — it gets predicate agreement in writing before you commit.

How many predicates can I cite in a 510(k)?

There is no formal limit on the number of predicates, but clarity matters more than quantity. One well-chosen primary predicate with 1-2 reference predicates is typical. Citing many predicates can dilute your argument and confuse reviewers. Be selective and ensure each predicate serves a clear purpose in your SE argument.

Can I use a 510(k)-cleared device from another country's clearance as a predicate?

No. A predicate must be a device legally marketed in the United States. Foreign regulatory clearances (CE marking, TGA listing, etc.) do not create valid predicates for FDA 510(k) submissions. The device must have a US 510(k) clearance, De Novo authorization, or pre-amendment marketing status.

What is the difference between a predicate device and a reference device?

A predicate device (primary or reference) is a legally marketed device to which you claim substantial equivalence. Each predicate must share the same intended use as your device. A "reference device" (as used in the FDA's SE guidance) is a broader term that can include devices cited to support scientific methodology or standard reference values at Decision Point 5 of the flowchart — these devices do not need to have the same intended use but must have a scientific rationale for their use.

Can a PMA-approved device serve as a predicate?

No. PMA-approved devices (Class III devices approved through the Premarket Approval pathway) cannot serve as predicates for 510(k) submissions. Predicates must be devices cleared through 510(k), authorized through De Novo, or marketed before May 28, 1976 (pre-amendment devices).

Can a device that was downclassified from Class III serve as a predicate?

Yes. A device that was originally classified as Class III and was subsequently downclassified to Class I or Class II can serve as a predicate for a 510(k) submission. Once the device is reclassified and is no longer subject to PMA requirements, it becomes eligible as a predicate — provided it was legally marketed.

Can a 510(k)-exempt device serve as a predicate?

Yes. A device that is exempt from 510(k) requirements can still serve as a predicate for another device's 510(k) submission, as long as it was legally marketed. The 510(k) exemption means the device itself does not need a 510(k) clearance to be marketed — but it does not prevent other manufacturers from referencing it as a predicate.

What is a "reference device" at Decision Point 5 of the SE flowchart?

The FDA's SE guidance uses the term "reference device" at Decision Point 5 to describe a device cited solely to support the scientific validity of performance test methods — not to establish SE. Unlike a predicate, a reference device at this point does not need to share the same intended use as your device. For example, you might reference a device from a different clinical application to demonstrate that a particular bench test method has been previously accepted by the FDA as valid for evaluating a specific performance characteristic. This is distinct from a "reference predicate" used to support the SE argument.

Does predicate selection affect which 510(k) submission type I should use?

Yes, indirectly. If your predicate is your own previously cleared device and the modification does not change the intended use or fundamental technology, you may qualify for a Special 510(k) — a faster pathway with a typical 30-day review. If your device type has well-established FDA guidance documents and recognized consensus standards, an Abbreviated 510(k) may be appropriate. For all other situations, a Traditional 510(k) is the standard path. See the section on Predicate Selection and 510(k) Submission Types above for details.

Key Takeaways

Predicate selection is a strategic decision, not a checkbox. It determines your entire submission strategy — testing requirements, comparison approach, and likelihood of clearance. Invest time in getting it right.
Intended use match is the first and most critical gate. If your intended use does not match the predicate's, nothing else matters. The FDA will issue an NSE determination.
Search systematically across multiple FDA databases. Use the 510(k) database, Product Classification database, GUDID, MAUDE, and the recall database. Document your search thoroughly.
Split predicates are prohibited. Every predicate cited in your submission must share the same intended use as your device. You cannot borrow intended use from one device and technology from another with a different intended use.
Prefer recent predicates with public summaries. They provide better information for planning your submission, reference current standards, and face less reviewer skepticism.
Check your predicate's safety history before committing. Search MAUDE and the recall database. Avoid predicates with design-related recalls or unmitigated safety issues.
Use a Pre-Submission (Q-Sub) when predicate selection is uncertain. It is free, and the FDA provides written feedback. This is the highest-value risk mitigation tool available for predicate selection.
Document everything. Your predicate search methodology, candidate evaluation, and selection rationale should be included in your submission. Proactive documentation strengthens your SE argument.
If no predicate exists, do not force-fit a 510(k). The De Novo pathway exists for exactly this scenario. A forced 510(k) with an inappropriate predicate wastes time and money.
Consistency is critical. Your intended use statement must be identical across every section of your submission — cover letter, Form 3881, device description, comparison table, and labeling. Inconsistency is one of the most common RTA triggers.