FDA publishes two dates for every food recall: when the recall was initiated, and when it appeared in the Enforcement Report. It never subtracts them. This page does, for every one of the … recalls in this atlas.
What this is not
This is not how long the public was unaware of a recall. The Enforcement Report is an administrative record. Firms routinely issue press releases, and FDA posts safety alerts, before that record publishes — this project holds examples of exactly those pages. A 40-day interval does not mean 40 days of silence.
What it does measure is how long the authoritative, structured, queryable record takes to exist. That is what anyone studying recalls after the fact depends on, and it is the record this entire atlas is built from.
How long recalls take to appear
Half of all recalls appear in the Enforcement Report between one and two months after they began.
Has it changed since 2017?
Median days by the year the recall was published. Bars are medians; the thin line through each is the 10th-to-90th percentile range, so a tall line means the year was inconsistent rather than merely slow.
Does severity make it faster?
Class I means a reasonable probability of serious health consequences or death. Class III is unlikely to cause harm.
How long recalls stay open
From initiation to the agency terminating the recall, for recalls it has closed out. Recalls still listed as Ongoing are excluded — they have no end date yet, and treating today as one would invent a number.
The longest waits
The ten longest intervals in the dataset. Every row links to the agency record it came from.
Method
- Both dates are published by FDA. The subtraction is ours, and is the only derived value on this page.
- Every record with usable dates is included. None are missing a date and none were published before they were initiated.
- One record is excluded. Its initiation date reads
0212-12-07— plainly a typo in the agency's data for 2012, giving an interval of 1,800 years. It is not corrected: inferring what the agency meant is not something this project does. It is stored verbatim, shown verbatim on its own record page, and left out of these statistics. The exclusion count appears below. - Medians are used rather than means throughout, because the distribution has a long tail — a handful of multi-year outliers move a mean and not a median.
- Recalls still Ongoing are excluded from "how long recalls stay open" only.
- This covers FDA-regulated food. USDA FSIS recalls — meat, poultry and egg products — are absent from this atlas entirely, so nothing here describes them.