A PMA withdrawal pays back. A 510(k) doesn't.

A sponsor pulls a submission mid-review because the data didn’t come together, or because a faster path opened up somewhere else. What happens to the user fee it already paid depends entirely on which pathway it filed under — and the gap between the two outcomes is bigger than most fee schedules let on.

What the statute actually provides

Section 738 of the Food, Drug, and Cosmetic Act — 21 U.S.C. § 379j — gives FDA limited, specific authority to refund a portion of a device user fee on withdrawal. Under § 379j(a)(2)(D), an application, report, or supplement withdrawn before FDA makes its filing decision gets 75 percent of the fee back. Withdraw after filing but before FDA’s first substantive action, and the refund becomes discretionary — FDA “may refund” some or all of the fee, based on how much review effort has already gone into the file. Withdraw after that, and there is nothing left on the table; the statute doesn’t ask the question, let alone answer it with money back.

That 75 percent figure is real, and it is exactly what a PMA, an HDE, or a device or biologics license application can draw on, under FDA’s own implementing guidance, User Fees and Refunds for Premarket Approval Applications and Device Biologics License Applications.

What the 510(k) guidance says instead

The parallel guidance for premarket notifications — User Fees and Refunds for Premarket Notification Submissions (510(k)s), issued by CDRH under docket FDA-2003-D-0198 and current in its October 2022 edition for MDUFA V — states plainly that the statute’s refund authority was never extended to a 510(k). A withdrawn 510(k) gets nothing back, whether the sponsor pulls it the day after paying the fee or six months into review. The guidance says why: by the time a submission reaches the point where a sponsor decides to withdraw, FDA has typically already committed real review resources to it, and Congress didn’t write a 510(k) refund into the statute the way it wrote one for an application.

A 510(k) fee comes back in exactly two situations, and neither one is a change of mind.

FDA determines the product at issue is not a device, or is a device exempt from 510(k) review under its own classification regulation — meaning the fee was never properly owed in the first place. Or the fee was submitted under the wrong fee type, or paid twice. Decide three weeks into review that the predicate doesn’t hold up, and the fee paid to find that out is simply gone.

The practice

The asymmetry runs opposite to the pathways’ usual weight: a PMA, the costlier filing with the longer review clock, carries a built-in early-exit discount that a 510(k) does not. A small RA team deciding whether to file a borderline submission and see what FDA says, versus doing more work up front to be certain, is pricing two different outcomes depending on which track it’s on — a decision that belongs with whoever owns the submission budget, not with any database. None of this is a field in a public record; it is a reading of the statute and the guidance built on it. This is regulatory intelligence and method, never regulatory advice about whether a specific submission is worth the risk of its own fee.