What eSTAR changed about writing a 510(k), not just filing one

Since October 1, 2023, a Traditional, Special, or Abbreviated 510(k) — along with its supplements and amendments — has had to arrive at FDA as an eSTAR submission rather than a free-form PDF, under the electronic-format authority Congress gave FDA at Section 745A(b) of the FD&C Act. Most of what got written about that shift at the time focused on the mandate itself: submit this way, or the submission isn't accepted for filing. The more consequential change is quieter, and it shows up before a sponsor ever clicks submit — eSTAR isn't a PDF form with required fields bolted on. It's a guided, branching interview that gates its own sections, and a submission author now makes decisions in an order the template enforces, not the order a table of contents used to suggest.

A form that asks the next question based on the last answer

eSTAR is built on FDA's PDF forms platform, structured around the same content the Traditional, Abbreviated, and Special 510(k) acceptance checklists already enumerate — device description, predicate comparison, performance testing — but instead of presenting them as headings in a document a submitter fills in from memory, the template walks the sections in sequence and, in several places, branches. An answer about whether the device includes software opens a software-documentation branch that a hardware-only device never sees at all; an answer about sterilization method opens the corresponding validation-evidence branch instead. A submitter who used to keep one master template and delete whatever didn't apply is now answering a question that decides which sections exist for this specific device, rather than pruning a document built to assume everything might.

It automates presence, not adequacy — the same line the acceptance review draws

eSTAR enforces that a required field can't be left blank before the file can even be assembled into a submission, which pushes a real category of “is this here” failures earlier — into authoring, before a Lead Reviewer ever opens the file. What it doesn't do is evaluate whether the content in that now-mandatory field is any good. That's the same limit that governs the Refuse-to-Accept review itself: a bench-test summary that's present, gated behind a required field, and satisfied with three sentences of boilerplate clears eSTAR's own gate exactly as cleanly as a page of rigorous data would. The template moved the presence check earlier in the process. It never touched the adequacy question at all.

De Novo caught up two years later, on its own clock

eSTAR's mandate for 510(k)s took effect October 1, 2023, under the final guidance FDA issued that year, updating the template it had first issued in September 2022. De Novo classification requests kept running on eSTAR voluntarily for two more years before the same electronic-format requirement caught up to them on October 1, 2025. A sponsor's authoring workflow built only around the 510(k) branch of the template — assuming a De Novo request could still be drafted the old way — would have needed rebuilding by that date, the exact kind of drift a mapped resource has to watch for rather than assume settled once and left alone.

What this means for a completeness checklist built for the old format

An internal 510(k) checklist built section by section against FDA's acceptance checklists, the way a working completeness checklist should be, still does real work under eSTAR — it's the tool that tells an author what evidence to have ready before the template asks for it, and which branch a given device's answers are going to open next. What that checklist can retire is any step describing how to lay the document out; eSTAR fixes the layout now. What it can't retire is the judgment about whether what goes into each field is actually sufficient — that question just moved into a different, guided container. eSTAR was never built to answer it.

A 510(k) preparation checklist built to track what eSTAR requires field by field — kept deliberately separate from any claim about whether the content in those fields is adequate — is previewed in the launch catalog. If your team has hit a branch this doesn't map correctly, the shelf takes that correction directly.