The firm files the report; FDA assigns the class

A correction or removal that clears 21 CFR 806.10's reporting bar generates one clear deadline: file with FDA within 10 working days of initiating the action. What that report doesn't do is assign the action a severity. Whether a correction or removal becomes a Class I, Class II, or Class III recall is a separate determination, made afterward, by FDA itself — not proposed, and not self-certified, by the firm that filed the report. 21 CFR Part 7, Subpart C runs that determination through its own mechanism, a health hazard evaluation conducted by FDA's own scientists, and a recall file that treats the 806 report as the end of the regulatory process has closed the file before the question Part 7 actually asks has even been reached.

Two sections, two different questions

This blog has already traced how 806.10's reportability call gets made — whether a correction or removal was initiated to reduce a health risk or remedy a violation that may present one, decided against the firm's own reasons and dated from initiation, not discovery. What happens after that report lands is a separate mechanism a recall file often assumes is already finished. Part 7 Subpart C isn't asking whether to tell FDA. It's asking how serious FDA itself judges the underlying hazard to be, on FDA's own scale, and that question doesn't get answered by the act of reporting.

The evaluation is FDA's, run by FDA's own people

21 CFR 7.41 has FDA conduct a health hazard evaluation, typically through an ad hoc committee of its own scientists, or a single FDA physician for a straightforward case. That evaluation weighs whether any disease or injury has already resulted, whether an exposed population is actually at risk and how large it is, the seriousness of the health consequences involved, and how likely those consequences are, immediately or over the longer term. A firm's own health-hazard framing, wherever it appears in its records or its 806 report, feeds that evaluation as input. It doesn't substitute for it, and a firm confident in its own severity assessment still gets FDA's independent one before the recall carries an official class.

Three classes, sorted by consequence, not by intent

21 CFR 7.3 defines the three outcomes plainly. Class I is a situation with a reasonable probability that use of, or exposure to, a violative product will cause serious adverse health consequences or death. Class II is a situation where use or exposure may cause temporary or medically reversible adverse health consequences, or where the probability of serious consequences is remote. Class III is a situation where use or exposure is not likely to cause any adverse health consequence at all. None of the three turns on how the problem arose, or how well-intentioned the firm's own 806 report was — only on what FDA's evaluation concludes the exposed population is actually facing.

The firm's job is a strategy, not a class

21 CFR 7.42 gives the firm a role in the process, but a narrower one than classification: developing a recall strategy tailored to the circumstances, weighing factors that include the results of the health hazard evaluation, how easily the product can be identified and traced, and how far down the distribution chain the recall needs to reach. FDA reviews and can modify that strategy, but the strategy is the firm's proposal for executing a recall FDA has already sized by class. A firm that drafts its public notice or plans its depth of recall — consumer level, retail level, wholesale level — before classification lands is planning around a severity it hasn't been told yet, and frequently has to redo that planning once the class is assigned.

The record needs two dated entries, not one

A recall file that only logs the date the 806.10 report went out is missing the entry Part 7 generates on its own timeline, often weeks later, and outside the firm's control entirely. The two dates measure two different decisions — the firm's own call that a report was owed, and FDA's own call about how severe the thing being reported actually is — and a file that collapses them into one status field loses the distinction the two regulations were written to keep separate.

Where this meets the rest of the file

The pattern that eventually triggers a correction or removal is frequently the same signal a CAPA tracker is built to catch first, as a complaint trend or a nonconformance rate rather than a single event — which makes the eventual health hazard evaluation a downstream consumer of records a quality system was already keeping for an entirely different reason.

A recall-tracking record built around this structure — the 806.10 reportability decision, kept visibly separate from FDA's own Part 7 classification once it arrives — is previewed in the launch catalog. If your program tracks these as one event rather than two, the shelf takes that correction directly.