Which hazards to test for in US food imports, by origin and product

This guide is for laboratories, certification bodies and QA teams who need to decide what to test a food for before it ships to the United States. It gives you a lookup table of the 15 origin-and-product lines with the most US hazard refusals since 2020, with the charges FDA brought in each and how often. Then it shows how to turn a row into a test plan, and where the table will mislead you.

The data is FDA's record of import refusals: shipments offered for entry that FDA refused. 27% of them are paperwork, such as a missing label panel or an unregistered facility. No lab test fixes those, so this table counts hazard refusals only: 18,793 of the 25,892 refusals from 1 January 2020 to 10 September 2026. The paperwork refusals in each line are shown in a separate column.

The table

Shares are the part of the line's hazard refusals that carried each charge. One refusal can carry several charges, so shares in a row can add up to more than 100%.

The linked product pages show the last 12 quarters and include paperwork, so their totals differ from the ones here.

How to use the table

Step 1: Find your line, not just your country. Mexico appears five times with different leading charges: colour additives in snacks, pesticides in fresh vegetables, filth in fruit. India's rice is almost all pesticides (98%), while its spices are mostly Salmonella (63%). Test for the product's hazard, not the country's reputation.

Step 2: Take the top charge as your baseline test. Where one charge covers most of a line, it tells you what FDA looks for and finds. Mexican snacks were refused for an unsafe color additive in 98% of hazard refusals. A microbiology panel on those chips answers the wrong question.

Step 3: Take filth seriously. Filth is the most common charge in seven of the 15 rows. It covers insects, rodent contamination, decomposition and foreign matter. If your test plan is all chemistry and microbiology, it does not cover the leading US charge for tamarind, jujubes, Indian bakery products and Mexican chillies.

Step 4: Look at the second and third charges. Mexican "other fruit" is split: filth 40%, pesticide residue 31%, Salmonella 25%. Any one test alone misses most of that line's refusals.

Step 5: Read the paperwork column before you quote a price for testing. China's "other vegetables" line had 88 refusals with no hazard charge at all, on top of its 556 hazard refusals. For those, the fix is the label or the registration, not the lab.

Worked example: spices from India

A spice exporter asks your lab what to test before shipping ground black pepper and blends to the US.

1. The row shows 538 hazard refusals since January 2020, and 261 of them came in 2024 or later. This is a current problem, not an old one.
2. Salmonella was charged in 63% of them, and filth in 40%. These are the two tests that matter.
3. An unsafe color additive was charged in 9%. That is small, but it is a separate test and a separate failure. Add it for coloured blends, not for whole pepper.
4. Only 24 refusals in this line were paperwork-only, so the lab plan covers most of the exposure.
5. Plan: Salmonella on every lot, a filth and extraneous-matter examination on every lot, and a colour screen for blends. The Salmonella page ranks every origin for the same charge, if the buyer is weighing another source.

Pitfalls

• The table covers under a third of the data. These 15 lines hold 5,908 of the 18,793 hazard refusals (31.4%). There are 1,220 origin-and-product lines with at least one. For anything else, open the product page or the hazard pages.
• It shows what FDA caught, not a full risk assessment. A hazard FDA does not look for never becomes a refusal. Use the table to cover what gets caught, then add what your own hazard analysis requires.
• A refusal is not always a lab result. The law lets FDA refuse a shipment if it appears, "from the examination of such samples or otherwise", to be adulterated. A refusal can follow an import alert rather than a test of that shipment. Products on an FDA import alert can be detained without physical examination, and the importer usually needs lab results to get them released. Melamine, which leads China's bakery row, is the subject of an FDA import alert for Chinese milk-containing products. For products like these, the lab report is what gets a shipment released, so this is where your work counts most.
• A charge name is not an analyte. In Indonesian shrimp, "insanitary handling" leads at 67%. But 323 of the line's 422 hazard refusals came from one processor, PT. Bahari Makmur Sejati, and 293 fell in September 2025. In August 2025 FDA had found the radioactive isotope caesium-137 in frozen shrimp from that firm and put it on an import alert. A filth test would not have found it. Read the refusals on the product page before you design a test around a charge name.
• FDA does not name the pesticide. The charge is "pesticide residue", with no active substance. Use a broad multi-residue screen rather than a targeted one.
• Old lines fade. Indonesian herbals had 290 hazard refusals since 2020, almost all kratom, but only 12 since January 2024. Check the product page's quarterly chart before you build a plan on history.
• Low counts mean no evidence either way. Of the 1,220 lines, 763 (63%) have fewer than five hazard refusals. A short record can mean clean product, low volume or little sampling.

Set up a watch

The table changes as refusals arrive. Set up a watch for Salmonella, and you will get at most one email a week when FDA publishes new Salmonella refusals, from any origin.