The 510(k) process, end to end.

1Phase 1 · Strategy

Strategy is the first phase because everything else compounds on the decisions you make here. The next four stops are about narrowing down: which pathway, which classification, which predicate, and how you'll argue your device is substantially equivalent to it. Get any of these wrong and the next six months get expensive.

Do you even need a 510(k)?

Not every device takes the 510(k) road — confirm your pathway before you spend a dollar.

The first decision is which FDA premarket pathway applies. There are four options: 510(k), De Novo, PMA, or none (your device is exempt). Get this wrong and you can spend months building the wrong submission.

A 510(k) is required when you're introducing a Class II (and a small set of Class I and III) device into commercial distribution, or modifying a legally marketed device in a way that could significantly affect safety or effectiveness. Foreign manufacturers can submit directly to FDA but will need a U.S. Agent as their point of contact.

Where people get stuck: assuming a 510(k) when there's no valid predicate (that's actually a De Novo), or building a full submission for a device that's already exempt.

Classify your device

Find your product code and regulation number — they define everything that follows.

Once the pathway is set, medical device classification gives your device the three labels that will follow it through every step: a product code, a regulation number (21 CFR), and a class (I, II, or III). Class determines pathway; product code determines which FDA guidance documents and special controls apply.

FDA wants a clear match to an existing classification regulation — and reviewers can spot a forced fit instantly. Picking a product code that's convenient rather than correct sets up a predicate mismatch later, which is one of the most common roots of a Not Substantially Equivalent (NSE) decision.

Find your predicate(s)

Your entire submission stands on the predicate you choose.

Everything in a 510(k) hinges on the predicate. A predicate device is a legally marketed device you'll compare yours to: a previously cleared 510(k), a granted De Novo, a pre-amendments device, or a reclassified one. The bar is that it has the same intended use as your device and hasn't been pulled from the market for safety reasons.

How do you find a predicate device? Most teams start in the FDA 510(k) database, filtering by product code and reading recent clearance summaries to find devices with closely matching intended use and technology. You can lean on a single primary predicate, multiple predicates, or a reference device to support a specific characteristic — but the more you stack, the harder reviewers have to work to follow your argument.

Where teams stumble: choosing a recalled predicate without checking, picking one with a subtly different intended use, or piling on so many predicates that the SE argument loses its anchor.

Set your substantial equivalence (SE) strategy

SE is the whole game — same intended use, and same or defensibly different technology.

Substantial equivalence is the legal standard for 510(k) clearance. It's a deceptively short rule with a lot of subtext: your device must have the same intended use as the predicate, and either the same technological characteristics or different ones that don't raise new questions of safety and effectiveness — backed by data.

SE strategy also drives which of the three 510(k) types you'll file:

• Traditional 510(k) — the default for new devices or significant modifications.
• Special 510(k) — for modifications to your own previously cleared device when design controls produce reliable results. Faster review (often 30 days).
• Abbreviated 510(k) — when you can rely on FDA guidance documents, special controls, or consensus standards instead of full head-to-head testing.

The output is an SE comparison table: a side-by-side mapping of every relevant characteristic of your device to the predicate's. Reviewers spend more time on this table than on any other section, and it's where most second-round questions originate. A technological difference that quietly raises a new safety question means more testing. Sometimes it means a kicked-back submission.

2Phase 2 · Evidence

Phase 2 is where you generate the evidence FDA will read. The risk analysis decides what testing you actually need; the testing itself is usually the longest and most expensive line item; and a Pre-Sub meeting — if you use it — lets you sanity-check your plan with FDA before you commit.

Risk & hazard analysis

Map the hazards now; they decide what you'll have to test.

A 510(k) needs a structured risk analysis (per ISO 14971) identifying hazards, the harms they can cause, and the controls built into the design. FDA doesn't dwell on the risk file the way they dwell on the SE table — but the risk analysis is the document that decides your testing scope. Treat it as paperwork and you'll discover missing tests halfway through Phase 2.

Risk analysis also feeds your QMS. If you're ISO 13485 certified or building toward FDA's Quality Management System Regulation (QMSR, effective February 2026), the same hazard analysis underwrites both your design controls and your 510(k) submission requirements.

Testing & performance data

The longest, most expensive phase — and the one to plan first.

This is the longest, most expensive phase of a 510(k) — and the one to plan first. Depending on your device you'll need some mix of:

• Bench / performance testing
• Biocompatibility (ISO 10993) for any patient-contact device
• Electrical safety and EMC (IEC 60601) for electromedical devices
• Software (IEC 62304 + FDA software guidance) for any device with software, including SaMD (Software as a Medical Device)
• Sterilization and shelf-life for sterile or implantable devices
• Human factors / usability for non-trivial UIs
• Cybersecurity for connected devices
• Clinical data, occasionally — when bench can't fully bridge to the predicate

If you're filing a 510(k) for an AI/ML-enabled medical device, this is also where you scope your Predetermined Change Control Plan (PCCP) — FDA's framework for pre-authorizing future model updates without filing a new submission for each one.

Running tests in the wrong order, missing a standard, or under-scoping all mean re-testing. Re-testing means time.

Pre-Submission (Q-Sub)

A free conversation with FDA that can save you a review cycle.

The FDA Pre-Submission program (Q-Sub) is a free way to get written FDA feedback on your predicate, test plan, or SE strategy before you submit. It's optional — but on novel or borderline devices it can save you a full review cycle later. Q-Sub requests use the PreSTAR template, FDA's interactive PDF for pre-submission interactions.

The trick is asking specific, well-framed questions. “Is this okay?” won't get you anything useful. “Here is our predicate and three alternatives — which would you support for an SE argument?” will.

3Phase 3 · Build

Phase 3 is the part most people picture when they think “510(k) submission” — drafting the document, packaging it into eSTAR, and paying the user fee. It's also where the strategy from Phase 1 gets tested against reality: inconsistencies between sections become very visible at this stage.

Draft the submission

Turn your strategy and data into the document FDA actually reads.

With strategy and data in hand, drafting means turning everything into the document FDA actually reads: device description, the SE comparison, proposed labeling and Instructions for Use (IFU), indications for use, and performance data summaries. The drafting itself is mostly assembly — the harder problem is keeping the sections internally consistent. An indication in the labeling that the testing doesn't support, or an SE comparison that contradicts the device description, is what reviewers catch and what triggers Additional Information (AI) requests.

Build the eSTAR package

The mandatory electronic format — and your best defense against rejection.

eSTAR is FDA's interactive PDF submission template, mandatory for most 510(k)s since October 1, 2023 (and for IVDs since 2024). Its built-in logic and completeness checks are why the old Refuse-to-Accept rejection pattern (see Stop 12) has collapsed. The mistake is treating the eSTAR template as a formatting step at the very end. Structure your content for it from the start: the package assembles cleanly when you do, and gets messy when you don't.

Pay the MDUFA user fee

FDA won't start reviewing until the fee clears.

Every 510(k) requires a Medical Device User Fee under MDUFA before FDA will begin review. The FY2026 510(k) user fee rates (Oct 1, 2025 – Sep 30, 2026):

Payment is by credit card. Allow a few days to clear before submitting or your package goes on “user fee hold.”

If you're eligible for small-business status, apply for Small Business Determination (SBD) before you file. It's free and saves you ~$19,500 on the 510(k) user fee.

4Phase 4 · Review

Phase 4 is FDA's turn. You upload the submission, get screened for completeness, and then a reviewer reads your work. Most of the 510(k) review time on a submission lives here — not because FDA is slow, but because Additional Information requests pause the clock while you respond.

Submit

Upload the eSTAR through the CDRH Portal — the clock starts.

Submissions go through the CDRH Customer Collaboration Portal. The MDUFA fee has to be paid and cleared before upload. The only surprises at this step come from leaving portal access until the last minute.

Acceptance review (technical screening / RTA)

A fast completeness check before the real review begins.

Once you're in, FDA does a fast technical screening to confirm the package is complete enough to review. For paper-era submissions this was the Refuse to Accept (RTA) checklist — a 15-day gate that bounced 30–40% of 510(k) submissions on the first try. eSTAR enforces completeness up front, so RTA screening is much lighter now, but it's still a real step that takes about 15 calendar days.

If FDA does issue an RTA hold, you have 180 days to fix the deficiencies and resubmit. Most RTA issues are administrative rather than scientific — missing sections, unsigned forms, fee not cleared.

Substantive review + Additional Information (AI) requests

The real review — and where most of the calendar time actually goes.

This is the real review. An FDA reviewer reads your SE argument and the data behind it. Within roughly 60 days you'll get a Substantive Interaction — FDA's first signal of where the review is heading. If they need more, they issue an Additional Information (AI) request — sometimes called a 510(k) deficiency letter — which puts the review on hold while you respond. You have up to 180 days to respond; the clock is on you, not them.

The biggest variable in your total 510(k) timeline is how fast and how completely you turn around AI responses. Slow or incomplete replies trigger second-round AIs that stretch the calendar by months and can push the submission toward an NSE decision.

The decision

Substantially equivalent, and you're on the market.

FDA closes the 510(k) review with one of three outcomes: Substantially Equivalent (SE) — you're cleared, you get a K-number, and you can market the device. Not Substantially Equivalent (NSE) — you can't market via 510(k); your options are De Novo, PMA, or a new 510(k) with a different predicate. Or you withdraw and resubmit. An NSE almost always traces back to Phase 1 — a weak predicate or an SE argument that wasn't right to start with. Which is why the first four stops matter most.