510(k) vs De Novo vs PMA: Choosing Your FDA Pathway

Once you know your product is a medical device under FDA's definition, the next decision, which of three pathways to pursue, shapes your timeline, your cost, and the specific evidence you need to build. This guide sets out how the choice is actually made, including a step-by-step decision tree you can walk through directly rather than a general description of the factors involved. For everything that follows the pathway decision, see the full FDA route for health software.

In short: The 510(k) pathway clears devices substantially equivalent to an existing predicate; De Novo classifies novel low-to-moderate-risk devices with no predicate; PMA is the full premarket approval route for high-risk Class III devices. Most health software targets a 510(k) if a predicate exists, De Novo if not.

The decision tree, step by step

This is a real branching flowchart, worked through as a sequence of numbered steps. Start at Step 1 and follow the logic that matches your device.

Step 1: Does a legally marketed device already exist with the same intended use and similar technological characteristics to yours?

• If yes → go to Step 2.
• If no → go to Step 3.

Step 2: Is that predicate genuinely close, meaning any technological differences are addressable with existing performance data rather than new clinical evidence?

• If yes → your pathway is 510(k). Proceed to predicate selection and a substantial equivalence comparison.
• If no, the closest available predicate is a weak match → go to Step 3, since a weak predicate often means you are better served treating the device as if no suitable predicate exists.

Step 3: Is your device high-risk, meaning Class III: it supports or sustains life, is implanted, or otherwise presents a potential unreasonable risk of illness or injury?

• If yes → your pathway is PMA, regardless of whether a predicate exists elsewhere in the same general category.
• If no, your device is low-to-moderate risk → go to Step 4.

Step 4: Would general controls, or general controls plus special controls FDA establishes for the new category, provide reasonable assurance of safety and effectiveness without a predicate to compare against?

• If yes → your pathway is De Novo classification. A granted request creates a new device classification and, in most cases, a new predicate category that later devices can cite.
• If no, even general and special controls look insufficient to manage the risk → your pathway is PMA.

Step 5 (run in parallel, at any point you're unsure): are you genuinely unable to tell whether a candidate predicate is close enough, or whether your device would be classified as novel or high-risk?

• If yes → file a 513(g) request for classification information before committing resources to a full submission strategy. This doesn't replace the steps above; it gets you a written FDA answer to plug back into Step 1 or Step 3 when the honest answer is "we don't know."
• If no → proceed with the pathway your Step 1 through 4 answers point to.

This ordering matters because the three pathways are not equally costly or fast, and misjudging predicate suitability early is the single most common reason a submission stalls or has to be restarted under a different pathway. A team that spends real engineering time on a device shaped around an assumed predicate, only to have FDA reject the equivalence argument during review, loses months it could have saved by running this tree honestly before design was locked.

Worked example: three devices, three answers

A continuous glucose monitor add-on that measures the same analyte, in the same way, as an existing cleared device, with a minor UI change, lands at Step 1 yes, Step 2 yes: 510(k). The technological differences are cosmetic enough that existing performance data plus a comparison table carries the argument.

A novel AI-based triage tool that flags a condition no cleared device currently screens for, using an algorithmic approach without a close predicate, lands at Step 1 no, Step 3 no (it's a decision-support tool, not life-sustaining), Step 4 yes: general and special controls, software validation, a clinical performance study, and labelling constraints are judged sufficient. This is a De Novo candidate, and several digital health devices have taken exactly this route to create the predicate category their competitors now use.

An implantable neurostimulator with a genuinely new mechanism of action, no existing predicate, and failure modes that could cause serious harm lands at Step 1 no, Step 3 yes: PMA, with the full clinical evidence burden that implies.

510(k): substantial equivalence and why predicate choice decides everything

The 510(k) pathway, named for the section of the Federal Food, Drug, and Cosmetic Act that establishes it, is a comparative review. You are not asking FDA to independently verify your device is safe and effective in isolation. You are demonstrating that it is substantially equivalent to a predicate device already legally on the US market, in intended use and, where technological characteristics differ, that those differences don't raise new questions of safety or effectiveness.

Predicate selection is the central strategic decision of a 510(k), not a formality completed after the real engineering work is done. A strong predicate shares your intended use closely and has technological characteristics similar enough that differences are straightforward to address with existing performance data. A weak predicate, chosen because it's the closest available match rather than a genuinely suitable one, forces you to carry the burden of proving that meaningful technological differences don't raise new safety or effectiveness questions. That's a harder and more evidence-intensive argument to win, and it's the scenario the decision tree above tries to route you away from at Step 2. Two further points harden a predicate choice: the predicate must itself be legally marketed in the US — one that was removed from the market for safety reasons, or later found not substantially equivalent, is a vulnerable foundation for your comparison — and "510(k)" is not monolithic: Traditional, Special, and Abbreviated 510(k) variants exist, with the Special 510(k) reserved for a manufacturer's modification to its own already-cleared device. Why Most 510(k)s Get an Information Request covers predicate mismatch as the leading cause of delay in detail.

De Novo: the novel-device route, honestly costed

De Novo exists for devices that are low-to-moderate risk but have no suitable predicate, typically because the underlying technology or intended use is genuinely new. Rather than comparing your device to something that already exists, FDA reviews it directly against general controls and, where FDA establishes them, special controls. If granted, the device is classified into Class I or Class II, creating a new device classification and, often, a new predicate category that subsequent similar devices can then use for their own 510(k) submissions.

The honest cost comparison with a 510(k): De Novo generally takes longer and requires more original evidence, because there is no existing comparator to lean on, and FDA is evaluating your safety and effectiveness case more directly rather than checking equivalence to a known quantity. It's also, in a specific strategic sense, more valuable to a first mover. A granted De Novo can become the predicate future competitors cite, a meaningfully different competitive position than being one of several devices citing a common, older predicate that everyone else already uses too.

PMA: what Class III really involves

Premarket Approval is FDA's most rigorous pathway, reserved for Class III devices, generally those that support or sustain human life, are of substantial importance in preventing impairment of human health, or present a potential unreasonable risk of illness or injury. PMA requires the most extensive evidence of the three pathways, typically including clinical data specific to your device rather than a predicate comparison or general controls, with a correspondingly longer timeline and higher cost.

This is also the only one of the three pathways where "approval" is the technically correct term. A 510(k) results in clearance; a De Novo results in classification, sometimes described as authorization; a PMA results in approval. This distinction reflects genuinely different legal standards of review, not a stylistic preference. Using "approved" loosely for a cleared 510(k) device is a factual inaccuracy that regulatory-literate readers, investors, and FDA itself will notice.

Pathway comparison at a glance

These figures are directional, not commitments — industry-cited planning context, not FDA-published statistics or a Venitara quote. Actual cost and timeline vary with device complexity, how much clinical evidence FDA expects for your specific risk profile, and how many rounds of questions your submission draws. The 510(k) row uses the same submission-to-clearance window and industry-cited cost range as Why Most 510(k)s Get an Information Request, which breaks down what each end of the range includes. Treat the table as a planning anchor for early budget conversations, not a quote.

The 513(g) option when genuinely unsure

A Section 513(g) request lets you formally ask FDA how it would classify your device, and by extension which pathway would apply, before committing to a full submission. This is narrower than a Pre-Submission meeting. It's specifically a classification question, and it produces a written FDA response rather than an informal consultation. It's most useful when genuine ambiguity remains after working through the decision tree above, for instance when a candidate predicate exists but its suitability is unclear, or when your device sits at a boundary between existing device categories. Founders sometimes treat asking FDA a direct question as a sign of weakness in a submission strategy. It isn't. A 513(g) response that confirms your assumed pathway is cheap insurance against restarting a submission months later, and one that corrects a wrong assumption is far cheaper to receive before you've built a design and evidence plan around the wrong pathway.

Pathway choice as strategy, not paperwork

The pathway decision is frequently treated as a late-stage administrative choice, made once the product is otherwise finished. This undersells how much it should shape earlier decisions. Predicate availability can inform product design choices made well before submission, since a device engineered with a clear predicate in mind from the start avoids the harder, more evidence-intensive De Novo argument later. Similarly, understanding early that your device is likely headed for a De Novo, because it's genuinely novel, changes your evidence-generation timeline and budget in ways worth planning for at the fundraising stage, not discovering after a 510(k) attempt has already stalled on predicate mismatch.

This is also where the decision tree earns its keep beyond a one-time exercise. Revisit it whenever your device's intended use, patient population, or core technology shifts meaningfully during development, since a pathway conclusion reached during early planning doesn't automatically hold if the product changes underneath it.

What actually drives cost and timeline within a pathway

The comparison table above gives ranges, but two devices on the same pathway can still land at very different points within that range, and it helps to know why before you budget. For a 510(k), the biggest swing factor is how much testing your predicate comparison requires beyond a paper equivalence argument. A device that's electrically and mechanically almost identical to its predicate might clear with bench testing alone. A device with a materially different user interface, a new sensor, or a software component the predicate didn't have will usually need additional performance testing, sometimes usability studies, and occasionally a small clinical component, each of which adds real weeks and real cost on top of the baseline.

For De Novo, the swing factor is usually how much original clinical evidence FDA expects to be satisfied that general and special controls manage the risk. A digital therapeutic with a well-established mechanism and a straightforward primary endpoint might need a single adequately powered study. A device making a more ambitious clinical claim, or intended for a vulnerable population, is more likely to need a larger or longer study, which is where a De Novo budget can move from "meaningfully higher than a 510(k)" to something closer to a scaled-down PMA.

For PMA, complexity and cost track the number and length of clinical studies FDA expects, plus manufacturing and quality system evidence that goes well beyond what a 510(k) or De Novo submission requires. A single-arm feasibility study followed by a pivotal trial is a common structure, and each stage adds time measured in years, not months, before a submission is even filed.

None of this changes which pathway you're on. It does mean two founders both told "you're on a 510(k)" or "you're on a De Novo" by an advisor should ask a follow-up question: given our specific technological differences and claims, where in that pathway's range are we likely to land, and what's driving it?

Frequently asked questions

Can we start with a 510(k) attempt and switch to De Novo if it fails? Yes, this is a recognised route: if FDA determines during 510(k) review that no suitable predicate exists, it can issue a "not substantially equivalent" determination alongside an invitation to pursue De Novo, effectively converting the submission. This is slower than correctly identifying the right pathway from the start, but it's not a dead end.

Does a granted De Novo mean competitors can never use a predicate for a similar device? No, the opposite: a granted De Novo typically creates a new device classification and predicate that subsequent similar devices can cite in their own 510(k) submissions, meaning competitors following you may actually have an easier 510(k) path than you did.

Is PMA ever the right choice for software-only devices? It's rare but not impossible, generally limited to software that directly controls or is inseparable from a Class III function, such as certain closed-loop therapeutic control systems. The large majority of standalone health software targets 510(k) or De Novo.

How long does a 513(g) request take, and does it commit us to anything? 513(g) requests generally take a few months for a written FDA response and don't commit you to any specific pathway. They're an information-gathering step, though FDA's response isn't binding in the way a formal submission determination is.

Does choosing the "wrong" pathway early cost us the ability to correct course? Not usually, but it costs time and money: a 510(k) submission built around a weak predicate, once FDA raises substantial equivalence concerns, often needs to pivot toward De Novo, which restarts much of the evidence-generation timeline rather than simply continuing from where the 510(k) left off.

If two different predicates seem plausible, how do we choose between them? Run the comparison against each candidate separately rather than picking the more convenient one first and hoping it holds up. The stronger predicate is usually the one with the closest intended use statement, not necessarily the one that's easiest to find or the newest device on the market. When the two candidates point to genuinely different arguments, a 513(g) request or a Pre-Submission meeting is a reasonable way to get FDA's read before you commit.

Can a single product line need more than one pathway for different features? Yes, this happens more than founders expect, particularly when a platform bundles a well-precedented core function with a genuinely novel add-on feature. In those cases it's worth evaluating whether the novel feature should be scoped as a separate submission, or whether it changes the classification of the combined product, rather than assuming the whole product inherits the simplest pathway of its component parts.

Does a foreign clearance or approval count as a predicate, or help our FDA submission? Not directly. FDA's substantial equivalence comparison looks at devices legally marketed in the United States, so a CE mark or another country's approval doesn't itself establish a predicate, though the underlying clinical or performance data generated for a foreign submission can sometimes support a US filing if it's relevant and well documented. Don't assume international market presence shortens the US pathway decision; it can inform your evidence package without changing which of the three pathways applies.

We think we're a 510(k), but nobody on the team has done one before. What's the actual first move? Before writing anything, search FDA's own database for the predicate you have in mind and read its clearance letter and, if available, its summary of technological characteristics. That single step either confirms your assumption or reveals a gap in your reasoning far more cheaply than discovering it after a submission is filed. If the predicate looks solid, a Pre-Submission meeting is a reasonable next step to confirm FDA's read before you invest in the full evidence package.

The free MedTech Compass can give you an AI-generated first read on which pathway likely fits your device, based on the inputs you provide. It's not a validated regulatory determination, and predicate research in particular benefits from expert review before you commit a submission strategy to it.

---

Where next: Does My Health AI Need FDA Clearance? · Why Most 510(k)s Get an Information Request · FDA PCCP: Updating AI Models After Clearance · Wellness App or Medical Device? The FDA Line

Find out in minutes which pathway likely fits your device. Start the MedTech Compass →