How to Find and Choose a Predicate Device for a 510(k) That Won’t Get Held Up

2026-09-16 · Caduvo Team

Predicate device selection determines the cost and timeline of your 510(k) submission. This guide covers how to search FDA databases, rank candidates, and avoid weak or split-predicate strategies that trigger extra testing.

Pick the wrong predicate device and your 510(k) runs straight into a hold letter, a surprise clinical-data request, or a year of back-and-forth with reviewers. Pick the right one and the testing burden shrinks, the review clock stays predictable, and your team can move on to the market-access problems that actually matter.

Substantial equivalence is the legal engine of a 510(k). The practical version is simpler: your new device needs to match—or come so close that the differences don’t raise new safety questions—a device that’s already legally on the market. That device is your predicate. Here’s how to find candidates, rank them, and avoid the strategies that blow up timelines.

What a Predicate Device Actually Means for Your 510(k) Submission

A predicate device is a legally marketed device to which you claim substantial equivalence. If FDA agrees, your device gets cleared to the same classification and regulation number—Class II, typically—without you having to prove safety and effectiveness from scratch.

Substantial equivalence has two prongs:

• Same intended use
• Same technological characteristics, or different technology but it doesn’t raise new questions of safety or effectiveness

That second prong is where teams get overconfident. If your device uses a new sensor material, a different energy source, or a novel algorithm, and the predicate didn’t, FDA will ask you to prove the change doesn’t add risk. That proof might mean bench testing, animal data, or a small clinical study. The predicate you choose sets the baseline for how much data you’ll need to generate.

Searching FDA Databases for Candidate Predicates

Don’t start by browsing competitor devices. Start with the product code.

Every device type FDA regulates has a three-letter product code and a regulation number. Your proposed device falls into one of those—maybe more than one if it combines functions. The product code dictates the classification panel, the recognized consensus standards, and the types of predicates that are reasonable to claim.

A practical workflow:

1. Go to the FDA Product Classification database and find the product code that best matches your device’s primary intended use. Write down the regulation number (e.g., 21 CFR 878.4400).
2. Open the 510(k) Premarket Notification database. Search by product code. Filter to the last 10-15 years—predicates older than that can still work, but a recent clearance signals reviewers are comfortable with the device type today.
3. Scan the “Device Name” column for devices that describe an intended use similar to yours. Download the 510(k) summary or statement for the top 5-10 candidates.
4. Read the Indications for Use statement in each summary. If it doesn’t match your own intended use, move on. FDA reviewers compare indication statements word-for-word.

Ranking Predicate Candidates So You Actually Send Fewer Tests

Not all matches are equal. Build a shortlist and score each candidate against a checklist:

• Same product code and regulation number? (Essential)
• Indications for use overlap with your device’s indication statement 100%? (Critical)
• Same fundamental operating principle? (e.g., ultrasonic vs. radiofrequency ablation—different principles mean you’ll need to bridge a new technology gap)
• Same materials in patient-contacting components? (If not, expect biocompatibility testing)
• Same clinical performance endpoints in the predicate’s 510(k) summary? (You’ll need comparable performance data)
• No history of recalls or safety signals linked to the predicate’s design? (A predicate with a troubled history can invite extra scrutiny)

The closer the match on every line, the heavier your reliance on non-clinical bench testing and the lighter your reliance on anything that involves an IRB. A near-identical predicate can let you skip clinical data entirely. A predicate that requires a “different technology” justification often means you’ll need performance data in humans.

The Risks Nobody Talks About: Weak and Split Predicates

A weak predicate is one that matches your intended use but has a different mechanism of action or relies on a different scientific principle. FDA will treat that as a new technology, even if the device looks similar on the shelf. The result: the review division asks you to run additional performance testing, sometimes at the last minute.

A split-predicate strategy is worse. That’s when a submitter points to device A for intended use and device B for technological characteristics. FDA will almost always reject it. The agency’s guidance is clear: you’re claiming substantial equivalence to a single device, not a collage. If your device truly has a new technology, use the De Novo pathway instead. Trying to stitch two predicates together is the fastest route to a refuse-to-accept letter.

The time to test your predicate choice is before you lock the design. Share your top candidate’s 510(k) summary with your testing lab and ask: if we claim equivalence to this device, exactly which standards and test protocols will apply? If the answer includes a clinical study you weren’t planning to run, you know the predicate is too far away.

Act on Your Predicate List, Don’t Just File It

A predicate is not a regulatory formality. It’s the single biggest lever for controlling your 510(k)’s cost, timeline, and outcome.

• Start with the FDA product code, not a competitor’s name.
• Audit at least 5-10 predicate summaries before narrowing to one.
• Rank candidates on indication match, technology match, and clearance recency—not sales volume.
• Reject split predicates outright.
• Validate your choice with the testing lab before the design freeze.

Caduvo’s platform assembles predicate clearance histories, product codes, and regulatory timelines so you can compare candidates without manually cross-referencing databases. When a few weeks of searching can eliminate a year of unnecessary testing, the right tool pays for itself in the first month.