How Long a 510(k) Really Takes (and Why Most Hit at Least One Round of Questions)

Founders planning around FDA's stated 90-day review clock are planning around a number that rarely reflects reality, and the preparation that moves it is set out in FDA clearance for health software. This guide breaks down what actually drives 510(k) timelines, why most applications face some form of hold or request for more information along the way, why that's different from outright rejection, and what "submission-ready" actually means in practice.

In short: A first-time 510(k) typically takes 6-12 months from submission to clearance. Industry-cited preparation and consulting costs span a wide range — from the low tens of thousands of US dollars for a simple software 510(k) up to roughly USD 150,000-350,000 for complex, consulting-heavy submissions — on top of FDA fees. The honest picture on friction during review is genuinely a range, not one clean number: by one widely cited industry estimate, roughly two-thirds of submissions hit some form of Refuse-to-Accept hold or Additional Information request at some point in the process, a broad "any friction" definition. Separately, and using a much narrower definition, final outright rejection rates are cited closer to 10-15%. Both figures are real; they're measuring different things, and this guide explains both.

The honest timeline: preparation, review clock, and holds

FDA's review clock for a 510(k) is 90 FDA days, and this is the number most commonly quoted. Two caveats sit behind it. First, the 90 days is a MDUFA performance goal FDA commits to meeting in most cases — not a guaranteed deadline it must hit for every submission. Second, and more consequentially for planning, it only measures FDA's active review time, and that clock pauses entirely whenever FDA places your submission on hold, most commonly through a Refuse to Accept (RTA) determination at the outset, or an Additional Information (AI) request during substantive review.

The realistic end-to-end timeline, from submission to clearance, is commonly cited at 6 to 12 months for a first-time submission. To be explicit about what that figure includes: it measures the window from the day you submit to the day FDA issues a clearance decision, and it already accounts for the fact that most submissions experience at least one hold cycle along the way — it is the same submission-to-clearance window used in the planning table in 510(k) vs De Novo vs PMA: Which FDA Pathway. Preparation time before submission, building the comparison to your predicate, compiling performance testing, and drafting the submission itself, is a separate phase entirely and isn't included in either the 90-day clock or the 6-12 month post-submission figure. A realistic project timeline from a standing start needs to add preparation time on top of the submission-to-clearance window, and for a team building its evidence base from scratch, that preparation phase is often the longer of the two.

Two different numbers, two different questions: friction rate vs. rejection rate

This is the section worth reading carefully before you repeat either number to an investor, a board, or a Notified-Body-experienced hire who will check it.

There are two genuinely different questions you can ask about 510(k) outcomes, and industry sources answer them very differently because they're measuring different things.

The first question: what share of submissions face some kind of friction, a hold, a request for more information, a delay, at any point before eventual clearance? By one widely cited industry estimate, the answer is roughly two-thirds of submissions, sometimes cited in the 64-69% range depending on the source and dataset. This figure counts an RTA at the outset, an AI request mid-review, or any other hold, regardless of whether the submission eventually clears. It is explicitly a "how often does the process generate at least one round of questions" number, not a "how often do applications fail" number.

FDA's own published data points the same direction. Per FDA's MDUFA V performance reporting (2nd Quarter FY2023 report), FDA issued an Additional Information request in the first review cycle for 63–68% of 510(k)s submitted in fiscal years 2018 through 2022. That's an FDA-published anchor for the roughly-two-thirds friction figure — with two caveats attached: it measures first-cycle AI requests specifically rather than "any friction," and those cohorts largely pre-date the mandatory eSTAR template, which industry commentary credits with reducing acceptance-stage holds since. FDA's current MDUFA performance reports and FDA-TRACK dashboards, published quarterly at fda.gov, are the right primary source for the latest cohorts.

The second question: what share of submissions are ultimately rejected outright, meaning they don't clear at all, or the applicant abandons or withdraws without ever reaching clearance? Figures for this narrower, final-outcome definition run substantially lower, commonly cited in the range of roughly 10-15%.

Both numbers appear in industry commentary, and both are legitimate answers to different questions. The mistake, and the one this article previously made, is treating the higher "any friction" figure as if it answered the "did it fail" question. It doesn't. The table below lays out the distinction explicitly.

We're presenting both numbers here deliberately, rather than choosing the more dramatic one for a punchier headline, because a founder, investor, or experienced regulatory hire who checks either figure should find the nuance already addressed here, not discover a competing number somewhere else and start wondering what other figures on this site are similarly oversimplified. Neither the 64-69% nor the 10-15% figure is itself an FDA-published statistic. Both are industry characterizations of FDA submission patterns, sourced from regulatory consultancies and firms that track submission outcomes, and methodology and underlying datasets aren't always fully disclosed in the original sources — though FDA's own MDUFA reporting, cited above, independently corroborates the roughly-two-thirds first-cycle AI-request rate for the FY2018-FY2022 cohorts. Treat the industry figures as directional estimates, not precise, audited numbers, and verify against a current primary source — FDA's latest MDUFA performance report or FDA-TRACK dashboard — before citing either one in a context where precision matters, such as investor materials or a board deck.

What this means directionally, and this part is consistent across the industry commentary we've reviewed, is that a clean first-pass review, no RTA, no AI request, is the exception rather than the rule. The realistic timeline for most submitters needs to build in at least one round of FDA questions as the expected case, not a worst-case contingency. At the same time, a hold or an AI request is not a sign your submission has failed or is likely to fail; the much lower final-rejection figure is the better guide to whether a well-prepared submission is likely to eventually clear.

The two dominant failure modes

Predicate mismatch is the most common substantive reason a 510(k) stalls. This happens when the predicate device selected doesn't sufficiently match the new device's intended use or technological characteristics, or when the differences between the two raise new questions of safety or effectiveness that the submission hasn't adequately addressed. Predicate selection isn't a formality. It's the central strategic decision in a 510(k), since your entire substantial equivalence argument is built on it, and a weak or poorly justified predicate choice is difficult to fully recover from mid-review without essentially restarting the comparative argument.

Documentation gaps cover a wide range of issues: missing or inadequately detailed performance testing, unclear or incomplete device description, software documentation that doesn't meet FDA's expected level of detail for the device's documentation level, or labelling that doesn't align with the claimed intended use. These are generally more mechanically fixable than a predicate mismatch, but they're also the more common reason for an RTA determination specifically, which happens at the very start of the process and effectively restarts your clock before substantive review has even begun. A related, entirely avoidable trigger sits at the format level: 510(k)s are now submitted through FDA's structured eSTAR template, and a technically incomplete or incorrectly assembled eSTAR package can itself hold up acceptance before anyone reads your substantive argument.

A third, less discussed pattern worth naming: internal inconsistency between sections of the same submission. A device description that says one thing, testing that was run against a slightly different configuration, and labelling that implies a third intended use, none individually wrong, but collectively confusing to a reviewer, is a common and entirely avoidable trigger for an AI request. This is less about any one piece of evidence being weak and more about the submission not reading as a single, coherent argument.

A worked timeline: what a hold cycle actually adds

Numbers describing a distribution across many submissions are useful for planning, but they can obscure what actually happens on a single project's calendar when a hold occurs. It helps to walk through a realistic case.

A first-time applicant submits a 510(k) for a moderate-risk connected diagnostic accessory. The predicate is reasonably well matched, and the performance testing is thorough, but the submission omits a specific piece of biocompatibility documentation FDA expects for the device's patient-contact material, an omission the team didn't realize was required because their predicate's public summary didn't detail it. FDA issues an RTA within the first 15 days, before substantive review even starts, because the omission is significant enough that FDA won't accept the file as complete. The clock resets. The team spends three weeks compiling the missing documentation, resubmits, and the file is accepted. Substantive review begins, and roughly 60 days in, FDA issues an AI request asking for clarification on one aspect of the performance testing protocol, a real but narrow question. The team responds within two weeks with a clarifying memo, no new testing required. FDA completes review shortly after and the device clears.

Total elapsed time from original submission to clearance: roughly seven months, comfortably inside the commonly cited 6-12 month range, despite two separate hold events. Neither hold reflected a fundamental problem with the device or the predicate; both were addressable gaps caught during review rather than before it. This is a fairly representative shape for what "hits some form of hold" looks like in practice: not a crisis, but two discrete delays a more thorough pre-submission review, or a Pre-Sub conversation with FDA, could plausibly have caught earlier.

Contrast this with a less fortunate scenario: a team selects a predicate that turns out to have a materially different intended use once FDA scrutinizes the comparison closely. The AI request here isn't a narrow clarification; it's a request to justify equivalence on a point the original submission didn't adequately address, because the predicate choice itself was the weak link. Responding requires new testing, not just a clarifying memo, adding several months, and in a worse version of this scenario, the gap is severe enough that the team ultimately withdraws and resubmits against a better-matched predicate. This is the failure mode a Pre-Sub is specifically designed to catch before it costs a full review cycle, and it's why predicate mismatch, not documentation gaps, is the more expensive of the two failure modes described below.

How a Pre-Sub de-risks the clock

FDA's Pre-Submission (Pre-Sub or Q-Sub) programme lets you request formal, written FDA feedback on key aspects of your planned submission, including your proposed predicate and testing strategy, before you commit to a full submission. This isn't a formal review and doesn't guarantee your eventual submission will clear without questions, but it substantially reduces the risk of discovering a fundamental predicate or evidence-strategy problem only after submission, when addressing it costs a full review cycle rather than a planning conversation.

Teams that skip the Pre-Sub step, usually to save the several months a Q-Sub round can add to the front of the timeline, are frequently trading a smaller, controlled delay early for a larger, less predictable delay later if their predicate or testing strategy turns out to have a fundamental issue FDA flags only during substantive review. Given how central predicate selection is to the entire submission, a Pre-Sub focused specifically on validating your predicate choice is often the single highest-leverage use of the programme, even for teams that otherwise feel confident in their evidence package.

Cost breakdown, attributed

Industry-cited figures for 510(k) preparation and consulting span a wide range, and it matters what each end of the range includes. A comparatively simple software 510(k) — clean predicate, limited testing burden, most drafting done in-house — is commonly cited in the low tens of thousands of US dollars. Complex, consulting-heavy first-time submissions are commonly cited at roughly USD 150,000 to 350,000, a figure that covers predicate research and comparison development, performance testing, and the submission drafting and compilation itself. Both ends are industry-cited planning context drawn from regulatory-consultancy benchmarking, not FDA-published figures or a Venitara quote. Neither end includes the cost of any additional testing required as a result of an AI request during review, which is a real and under-budgeted risk for submissions that go in with thin original testing. On top of preparation costs sit FDA's own user fees, set annually under MDUFA: qualifying small businesses (broadly, those with no more than USD 100 million in revenue) pay a substantially reduced fee, but must obtain FDA's small-business certification in advance of submitting — check FDA's current fiscal-year fee schedule for the exact amounts. Teams budgeting a complex submission should treat the top of the relevant range, plus a contingency for at least one round of additional testing, as the realistic planning number rather than anchoring on the low end.

What "submission-ready" means and why it's the metric that matters

Given that a clean first pass is the exception under either definition discussed above, the more useful question for a founder to ask isn't "will FDA accept this" in the abstract, but "is this submission built to materially improve the probability of a clean or fast review." Submission-ready, in practical terms, means a predicate selection with a documented, defensible rationale addressing likely points of comparison scrutiny, performance testing that anticipates the specific questions a reviewer is likely to ask given your device type and predicate, and a submission package internally consistent across device description, testing, and labelling, since inconsistencies between these sections are a common, avoidable trigger for AI requests.

This is deliberately framed as materially improving your probability of a faster, cleaner review, not as a guarantee of any specific outcome or timeline. No party, including Venitara, can guarantee an FDA clearance decision or its timing, and any claim otherwise should be treated with real scepticism. What a disciplined, submission-ready process changes is the probability distribution of outcomes, not the certainty of any single one.

Frequently asked questions

Does a Pre-Sub guarantee our eventual 510(k) will clear without an AI request? No. It substantially reduces the risk of a fundamental, late-discovered problem with your predicate or evidence strategy, but it doesn't eliminate the possibility of review questions on other aspects of the submission, and it isn't a guarantee of any outcome.

If we get an Additional Information request, does that mean our submission failed? No, and this is worth stating plainly because it's a common source of founder anxiety. An AI request is a normal, common part of the review process; by the broader "any friction" industry estimate discussed above, it affects most submissions in some form. It pauses FDA's clock while you respond, adding time, but it isn't a rejection of the submission, and the large majority of submissions that receive an AI request go on to clear. The much lower final-rejection figure, roughly 10-15% by the narrower definition, is the more relevant number if what you actually want to know is "will this eventually clear," rather than "will this be perfectly smooth."

Why does this article cite two different numbers instead of one? Because the two numbers answer two different questions, and presenting only the larger one without the definition attached is misleading. Roughly two-thirds facing some hold or request is one industry estimate of process friction; roughly 10-15% final rejection is a separate estimate of ultimate outcome. Both are directionally useful for different purposes: the first sets realistic timeline expectations, the second tells you that most submissions that enter the process eventually succeed.

How is a hold or AI request different from an outright rejection? A hold, whether an RTA at the outset or an AI request mid-review, pauses the clock and asks for something: missing information, additional testing, or clarification. It doesn't end the submission. An outright rejection or a withdrawal means the submission doesn't reach clearance at all. The two are measured by different figures precisely because conflating them overstates how often submissions actually fail.

Can predicate mismatch be fixed after an AI request, or does it mean starting over? It depends on severity. A narrow gap in the comparison argument can sometimes be addressed with additional data in response to an AI request. A fundamentally unsuitable predicate often can't be fixed within the same submission and may require withdrawing and resubmitting with a different predicate, which is the more costly outcome a Pre-Sub is specifically designed to help you avoid.

Does device software documentation level affect how long review takes? Yes. FDA assigns software-containing devices a documentation level (Basic or Enhanced) based on risk, and Enhanced documentation devices require substantially more detailed software documentation, which both takes longer to prepare and is more likely to generate detailed reviewer questions if it's incomplete.

Should we mention either figure, the two-thirds friction estimate or the 10-15% rejection estimate, to investors? If you do, cite the specific figure with its definition attached, the way this article does, rather than a bare percentage. An investor or diligence team with regulatory experience is likely to know these numbers are contested, and presenting one without its definition is the kind of imprecision that undermines credibility on other, more important figures in your materials.

Is it worth paying for a formal gap analysis before submitting, on top of a Pre-Sub? For teams with a novel device, an unusual predicate comparison, or a first-time regulatory team, generally yes. A gap analysis focused on internal consistency, checking that device description, testing, and labelling tell the same story, catches the kind of avoidable inconsistency that generates AI requests. It's complementary to a Pre-Sub rather than a substitute: the Pre-Sub tests your strategy against FDA's expectations, while a gap analysis tests your submission's internal coherence before it reaches FDA.

Submission-ready before you submit is the standard worth building toward, not a clean-review guarantee, which no one can honestly offer. A structured conversation about your specific predicate strategy and evidence base is the right next step if you're planning a submission in the next 6 to 12 months.

---

Where next: 510(k) vs De Novo vs PMA: Which FDA Pathway · Does My Health AI Need FDA Clearance? · QMSR: What Replaced FDA's Quality System Reg · CE Mark vs FDA Clearance: The Real Differences

Get submission-ready before you submit. Book an expert conversation →