2025's 41% jump in US food refusals: one shrimp alert, then a broad rise

US food import refusals rose from 3,984 in 2024 to 5,625 in 2025, up 41%. One Indonesian shrimp exporter, stopped at the border after FDA found cesium-137, accounts for 298 of the extra 1,641 refusals (18%). The rest of the rise was broad: without that firm, refusals still rose 34%, and 2026 is running at the 2025 level.

If you have seen the 2025 number and want to know whether it was one event or a real change: it was both. Between 2020 and 2024, refusals grew 3% to 18% a year.

One firm, one month

In 2024, FDA refused 65 food entries from Indonesia. In 2025 it refused 403. 298 of those were shrimp from a single processor, PT. Bahari Makmur Sejati, which had no refusals in 2024. 283 came in September 2025, 95 of them on 10 September alone. 284 of the 298 carry the charge that the product "may have been rendered injurious to health" by insanitary conditions.

That matches FDA's own record (not from the refusal data): on 14 August 2025 FDA added the firm to Import Alert 99-51 for cesium-137, covering all its shrimp products, and that alert's refusal charge is the same wording. From 31 October 2025 FDA also required import certification for shrimp and spices from Java and Lampung. Only 18 of the 298 refusals record an FDA sample analysis.

The event was short. Indonesia had 83 refusals in January to August 2026, 26 of them seafood.

The rest was broad

Without that firm, 2025 still shows 5,327 refusals, 34% more than 2024:

Of the 46 origins with at least 20 refusals in either year, 36 rose. All five FDA import divisions rose, from +21% to +67%. Every calendar month of 2025 was above the same month of 2024, even without the shrimp firm.

Repeat shippers still matter. Ten firms account for 42% of the rise, and nine firms had 40 or more refusals in 2025, against four in 2024. Unsafe colour charges nearly doubled, from 363 to 714. Mexico's share went from 99 to 244, 193 of them corn or potato chips, and 146 from just two firms (see Mexican snack foods).

Hazard or paperwork?

Mostly hazard. Refusals citing a safety hazard rose from 2,847 to 4,156; paperwork-only refusals from 1,137 to 1,469. Without the shrimp firm, hazard refusals rose 36% and paperwork 29%. By charge group, again without the shrimp firm, the biggest increases were:

One pattern stands out. Refusals that record an FDA sample analysis did not rise: 1,096 in 2024, 1,066 in 2025. The whole increase is in refusals with no sample analysis, which went from 2,888 to 4,559. So the jump did not come from more FDA lab findings. The data shows that shift; it does not show why. Outside the cesium alert we found no FDA announcement that explains it, and we do not guess at one.

It is not a data artefact, and it has not reversed

The FDA export has no duplicate lines, and entries per year in the raw file match the site's count exactly. Lines per entry (2.29 vs 2.32) and charges per entry (1.73 vs 1.67) barely moved, so entries are not being split or merged differently.

The new level held. January to August 2024 had 2,533 refusals; the same months had 3,578 in 2025 (none of them from the shrimp firm) and 3,594 in 2026. Filth and insanitary refusals keep climbing: 508, 841 and 975 in those three periods.

What this means if you…

• …import into the US: treat 2025 as the new baseline, not a spike. Most of the increase came without an FDA sample, so a clean lab result from your supplier may not be enough to avoid a refusal. Check whether your suppliers appear on an FDA import alert.
• …write about US food safety: "refusals up 41%" is accurate but needs two qualifiers. About 7.5 points of it are one cesium-137 case. The rest is more refusals without FDA sampling, not more contamination found by sampling.
• …run supplier QA: watch firm-level repeats. Nine firms passed 40 refusals in 2025. A firm on an import alert's red list has its shipments detained without physical examination, as the shrimp firm's 298 refusals show.

---

Method. 25,892 FDA import refusals dated 1 January 2020 to 10 September 2026 from FDA's Data Dashboard (human foods), one row per entry, dated by its earliest refusal line. Recent weeks may still fill in. "Sample analysis" is the export's own FDA Sample Analysis flag, counted per entry if any line carries it. Hazard vs paperwork uses the site's classifier: an entry counts as a hazard if any of its charges falls in a hazard group. An entry can sit in more than one charge group. Origin is the country FDA records for the entry. The shrimp firm is identified by the firm name in FDA's export.

Get an email when new filth and insanitary-conditions refusals appear: