510(k) vs De Novo: Which FDA Pathway Fits Your Device?
The single fact that decides which of these two pathways is even on the table is whether a legally marketed "predicate" device already exists that your product resembles. If a close-enough predicate exists, 510(k) — clearance by demonstrating substantial equivalence to that predicate — is generally the pathway a device follows. If no suitable predicate exists but the device is not high-risk enough to require the more demanding Premarket Approval (PMA) track, De Novo — a route for establishing a new category from scratch — is generally the one in play. That single fork explains most of the practical differences founders actually feel: how much of your evidence plan is comparative versus original, how your validation work should be sequenced, and how early you need a regulatory specialist in the room.
Nothing below tells you which pathway applies to your product, or whether your product is a regulated medical device at all. Those are fact-specific determinations that depend on your device's intended use, its technological characteristics, and the current state of FDA classification and guidance — determinations only qualified FDA regulatory counsel or a regulatory affairs specialist can make for your specific product. Treat this article as the vocabulary and decision framework you bring into that conversation, not a substitute for having it.
Quick Answer: 510(k) is generally used when a legally marketed predicate device already exists and your product can be shown to be substantially equivalent to it. De Novo is generally used for novel, low-to-moderate-risk devices where no suitable predicate exists. Both differ structurally from Premarket Approval (PMA), which generally applies to higher-risk devices and demands the most extensive evidence package. Which one actually fits your product is a determination for FDA regulatory counsel or a regulatory affairs specialist — not something to settle from a blog post.
Side-by-side: 510(k) vs. De Novo vs. PMA
Before you can even ask "which pathway," you need the vocabulary to have that conversation with a regulatory specialist without wasting their first two calls on definitions. The table below lays out the structural differences at a category level — not specific current procedures, fees, or timelines, which change and which you should confirm directly with FDA or counsel.
| Factor | 510(k) | De Novo | PMA (for context) |
|---|---|---|---|
| Predicate device required | Yes — a legally marketed device with the same intended use and comparable technological characteristics | No — used precisely because no suitable predicate exists | No predicate concept; evidence stands on its own |
| Core evidence question | "Is this substantially equivalent to an existing, legally marketed device?" | "Is this novel device's risk adequately controlled by special controls, without a predicate to compare against?" | "Does the totality of evidence show this device is safe and effective for its intended use?" |
| General risk-class fit | Typically lower- to moderate-risk device categories | Typically low- to moderate-risk, but novel enough that no predicate exists | Typically the highest-risk device categories |
| Relative evidence burden | Generally comparative — performance data measured against the predicate, not built from zero | Generally more original than a 510(k) — you are proposing the special controls that define the category | Generally the most extensive — usually requires the deepest clinical evidence package of the three |
| What a successful submission creates | Clearance for your device, referencing the existing predicate | A newly classified device type, which can itself become a predicate for later 510(k) filers | Approval specific to your device; does not create a new predicate category in the same way |
| Where cost and timeline data live | Pathway choice is only one input into cost and timeline — see our FDA clearance cost and timeline data for a fuller, sourced picture | Same — cost and timeline depend heavily on testing scope and consulting need, not pathway alone | Same |
Two things to notice in that table and nowhere else: the "predicate required" row is the whole ballgame, and the last row is a deliberate refusal to put specific numbers in a comparison table. Any article that hands you a specific dollar figure or day count for "the 510(k) process" without qualifying it by device type, testing scope, and current FDA fee schedules is giving you a number that will be wrong for your product in one direction or the other. Get current, product-specific figures from FDA's own fee schedules and from counsel, not from a general comparison table.
When 510(k) is generally the right conversation to have
510(k) tends to be the pathway founders end up discussing with regulatory counsel when their product looks, on a plain-English description, like something already legally on the market. That doesn't mean identical — it means comparable enough in intended use and technological characteristics that a reviewer could plausibly accept an argument of substantial equivalence, supported by performance data rather than a full standalone clinical evidence program.
The practical signal worth noticing, long before you talk to counsel, is whether you can point to a specific, named, legally marketed device and say "we do roughly what that does, for the same intended use, with a similar mechanism." If you can browse public FDA device databases and find something that looks like a real candidate predicate, that's a signal worth bringing to a regulatory specialist early — not a green light to assume 510(k) is your pathway. Predicate suitability is judged on intended use and technological characteristics together, and a device that looks similar on the surface can still fail a substantial-equivalence argument if its underlying technology or claims diverge in ways that raise different questions of safety or effectiveness.
This is also where the software question gets sharpest. A tool that takes patient-specific data and uses it to inform a specific diagnostic or treatment decision looks more like a medical device, generally speaking, than a tool that delivers general wellness or purely informational content with no patient-specific clinical claim. That distinction — general factors, not a determination — is the reason so many digital health founders end up needing a regulatory read early: the same feature set can sit on either side of that line depending on exactly how the intended use is described and marketed. If you haven't already worked through whether your concept is a device at all, our guide on how to validate a medical device startup idea walks through the clinical, regulatory, and reimbursement gates in sequence, with pathway scoping as one of the early stages rather than an afterthought.
When De Novo generally enters the conversation
De Novo generally comes up when a founder or their regulatory advisor searches for a predicate in good faith and comes up empty — the device's intended use, mechanism, or combination of the two is genuinely new enough that nothing legally marketed matches closely enough to support a substantial-equivalence argument, but the device also isn't so high-risk that it needs to go straight to the PMA track.
Structurally, De Novo asks a different question than 510(k). Instead of "is this equivalent to something that already exists," it asks "can this novel device's risks be adequately managed through a defined set of special controls, even without a predicate to lean on." That tends to mean more of the evidence burden sits on you — you're not borrowing credibility from an existing cleared device, you're proposing the framework that the device category will be judged against, potentially including the framework future competitors will use to cite your device as their predicate.
This is a common position for genuinely novel software-based tools, new sensor modalities, or first-of-kind combinations of existing technologies, precisely because the predicate landscape for anything truly new is thin or empty by definition. It is also, generally speaking, a pathway where working with a regulatory specialist matters more rather than less — proposing special controls for a device category that doesn't yet exist is a technical and legal exercise, not a checklist you complete from public guidance documents alone.
Edge cases: PMA territory and pathway switching
Two situations are worth naming explicitly, because they're exactly where founders get into trouble trying to self-navigate instead of engaging a regulatory specialist early.
You might be in PMA territory, not 510(k)-or-De Novo territory at all. The highest-risk device categories generally sit outside both pathways discussed here and require the more extensive PMA process, built around a fuller clinical evidence package rather than equivalence or special controls alone. If your device carries meaningful risk to patient safety — invasive components, life-sustaining function, or high-stakes diagnostic or treatment decisions made with limited human oversight — don't assume you're choosing between 510(k) and De Novo before confirming you're not actually looking at PMA. This is squarely a question for regulatory counsel, not a self-assessment.
Pathway switching happens, and it's expensive to discover late rather than early. A 510(k) submission can come back with a finding that the device is "not substantially equivalent" to the proposed predicate — at which point a founder who assumed 510(k) may find themselves needing to pursue De Novo (if the risk profile fits) or PMA instead, often after having already built and tested toward the wrong evidence package. This is the scenario the decision framework above exists to reduce, not eliminate: searching for a genuinely strong predicate early, with a regulatory specialist's help, is cheaper than discovering there isn't one after a submission. The same logic runs in reverse — a device planned as De Novo sometimes turns out to have a workable predicate a specialist finds that a founder's own database search missed.
Both edge cases point to the same operating rule: bring in FDA regulatory counsel or a regulatory affairs specialist before your engineering and evidence roadmap is locked to a specific pathway, not after a submission comes back with an answer you didn't plan for. Pathway choice is a strategic, evidence-shaping decision, not paperwork you attach at the end.
How pathway choice reshapes your validation plan
Whichever pathway your regulatory specialist ultimately confirms, the pathway decision itself changes what "validation" means for your device well before any submission is drafted. A 510(k)-shaped plan pulls your team toward finding and studying the right predicate, running comparative performance testing, and keeping your intended-use language close enough to the predicate's to support equivalence. A De Novo-shaped plan pulls your team toward building an original evidence case and toward defining, defending, and documenting the special controls that will justify the classification you're proposing — closer in spirit to building a case from first principles than to comparison.
That's why pathway scoping belongs early in your broader validation sequence, not as a regulatory afterthought once the product is built. The general discipline of pairing every risky assumption with the cheapest test that could disprove it — the same discipline behind the complete guide to startup idea validation — applies just as much to "will a regulator accept our predicate argument" as it does to "will a customer pay for this." Treat the pathway question as a testable, killable assumption you resolve with a specialist early, not a fact you discover by submitting and waiting.
Key Takeaways
- Predicate availability is the primary fork. A legally marketed device your product resembles closely enough points toward 510(k); no suitable predicate, combined with a low-to-moderate risk profile, points toward De Novo.
- 510(k) leans comparative; De Novo leans original. 510(k) evidence is generally measured against a predicate. De Novo evidence generally has to stand on its own and can define special controls that later become the predicate for others.
- PMA is a third, higher-risk track, not a variant of either. Don't assume you're choosing between 510(k) and De Novo before ruling out that your device sits in PMA territory.
- A "not substantially equivalent" finding can force a late, expensive pathway switch. Searching hard for a real predicate early — with a specialist — is cheaper than discovering there isn't one after submitting.
- Software founders face the sharpest version of this question. Whether a tool's use of patient-specific data to inform a clinical decision makes it a regulated device — and which pathway would apply — is a fact-specific call for regulatory counsel, not something to infer from feature similarity to an app you found.
- Never treat specific costs, timelines, or approval statistics from an article as settled. They vary by device type, testing scope, and current FDA fee schedules, and they change. Confirm current figures directly with FDA or your regulatory advisor, and see our FDA clearance cost and timeline data for a fuller, sourced treatment of that question.
- Bring in a regulatory specialist before the roadmap is locked, not after a rejection. Pathway choice shapes engineering and evidence decisions early enough that discovering it late is expensive to unwind.
Frequently Asked Questions
Can I just choose whether to file 510(k) or De Novo?
Not really — the pathway follows from your device's classification and whether a suitable predicate exists, not from founder preference. You can and should investigate the general landscape yourself (searching for plausible predicates, reading how similar products were classified), but the actual pathway determination, and whether your product is a regulated device at all, is made through the regulatory process itself, informed by qualified regulatory counsel. Treat "which pathway" as a question you resolve with a specialist, not a choice you make unilaterally.
What happens if FDA finds my 510(k) submission is not substantially equivalent to my proposed predicate?
Generally, a "not substantially equivalent" finding means the device cannot be cleared through that submission as filed, and the company has to reassess its regulatory route — which may mean pursuing De Novo if the device's risk profile fits, pursuing PMA if it doesn't, or revisiting the predicate argument entirely. The specifics of what happens next depend heavily on the device and the reasons for the finding, which is exactly the kind of fact-specific situation to work through with regulatory counsel rather than guess at generically.
Is De Novo generally more expensive or slower than 510(k)?
There's a structural reason De Novo submissions often involve more original evidence work than a straightforward 510(k) with a strong predicate — you're building the case for a new category rather than pointing to an existing one. But translating that into a specific cost or timeline comparison isn't something to take as settled fact from a general article, since both figures depend heavily on your specific device, testing scope, and consulting needs, and FDA fee schedules change over time. For a fuller, sourced treatment of cost and timeline questions, see our FDA clearance cost and timeline data.
Does my software product need to go through either of these pathways?
It depends on what the software actually does and claims — and that determination is not one this article, or any general article, can make for your specific product. As a general factor to be aware of, software that analyzes patient-specific data to inform a specific diagnostic or treatment decision tends to draw more regulatory scrutiny than software offering general wellness or informational content with no patient-specific clinical claim. That's a factor to bring to a regulatory affairs specialist early, not a self-service test you apply to conclude your product is or isn't a medical device. Our guide on validating a medical device startup idea covers where regulatory pathway scoping fits into your broader validation sequence.