Reading RASFF data without getting it wrong

This guide is for journalists, researchers and consultants who quote EU food safety data. It covers seven mistakes that turn a correct RASFF figure into a wrong sentence. Each one comes with a real example from the data, showing the wrong reading next to the right one. Read it before you write "country X has a food safety problem" or "rejections fell this year".

RASFF is the EU's Rapid Alert System for Food and Feed. The European Commission set it up in 1979 so that food safety authorities can exchange information on food and feed risks quickly. Member states file notifications, and the Commission publishes them in the public RASFF Window. This site reorganises that data by origin, product, hazard, notifying country and week. Unless stated otherwise, the counts below cover the 32,133 notifications from 23 January 2020 to 21 September 2026.

Mistake 1: treating a notification as proof the whole trade is contaminated

A notification is about one batch, container or product found at risk. It says nothing about the other consignments of the same product. Nor is every notification a serious risk. Each carries the authorities' own risk decision, and only 52% of notifications since 2020 are graded "serious". For border rejections the figure is 50%. Of the other half, 1,639 were graded "not serious" and 2,840 "undecided", "potential risk" or "potentially serious".

Detection also depends on how hard the EU looks. EU Regulation 2019/1793 requires identity and physical checks, including lab analysis, at a set frequency for listed origin-product pairs. Turkish and Iranian pistachios face 50% checks for aflatoxins. The site lists no such check for US pistachios. More checks find more failures, so part of a gap in counts can be a gap in scrutiny.

Checks do not create failures, though. Groundnuts from Argentina and from the United States face the same 20% aflatoxin checks, yet Argentina's rate is 4.7 per $1 billion of EU imports against 199.0 for the US.

Right reading: "X consignments were notified", not "X's product is unsafe". Say which risk decision the notifications carry, and whether the product is under increased checks.

Mistake 2: confusing the notifying country with the origin

The notifying country is the member state that found the problem. The origin is where the product came from. In 83% of notifications, the notifying country is not among the origins.

Wrong reading: "Bulgaria is the EU's biggest food safety offender." Right reading: Bulgaria borders Turkey, and much Turkish road freight enters the EU there, so it rejects a lot of Turkish goods. A member state that receives more of an origin's shipments will usually notify more of them.

Mistake 3: mixing the notification classifications

RASFF classifies each notification. The definitions are set in EU Regulation 2019/1715:

• Border rejection: a consignment refused at the EU's external border because of a risk.
• Alert: a serious risk that needs, or might need, rapid action by another member country.
• Information notification: a risk that does not need rapid action elsewhere. For follow-up means the product is or may be on another member's market. For attention means it is only in the notifying country, never reached the market, or is no longer on it.

A country's rank depends on which you count. Poland is the second most-notified origin overall, with 2,271 notifications, and third for alerts, with 798, of which 501 concern Salmonella. It has 3 border rejections, and ranks 96th on that count. That is because EU products don't cross the EU's external border. The Poland origin page counts border rejections and shows 3. The Salmonella page counts every classification and names Poland its most-cited origin. Brazil goes the other way: 4th for border rejections, 38th for alerts.

Right reading: always name the classification. "Most rejected at the border" and "most notified" are different lists.

Mistake 4: comparing raw counts instead of rates

A big exporter gets more rejections simply because it ships more.

Worked example: pistachios. Over the 12 quarters each page shows, Turkey had 66 border rejections for pistachios and the United States had 52. On counts they look similar. Measured against EU imports of pistachios from each, Turkey runs at 244.0 rejections per $1 billion and the US at 17.9. That is almost 14 times lower, and puts the US last of the three origins ranked. Iran had the most rejections, 86, but its page shows no rate. A missing rate means "not computed", not "low".

Right reading: compare rates only within one product. A pistachio rate and a fig rate are measured against different trade and cannot be ranked. For the same reason the site shows no rate for a country's food exports as a whole. And the rate is against trade value, not consignments inspected, so it is not a failure rate per shipment.

Mistake 5: comparing a partial year with a full year

On 21 September, 2026 held 3,859 notifications against 5,328 for all of 2025.

The naive reading says notifications fell. Like for like, they rose. Every "this year" total on the site, and the current week and quarter, is still filling up.

Right reading: compare the same calendar window, or use completed years only.

Mistake 6: forgetting that one notification can name several origins

1,219 notifications (3.8%) name more than one origin country, and 582 name none. The site counts a notification once for each origin it names. So origin totals add up to 33,049, more than the 32,133 notifications. Take notification 2026.7068, filed by Czechia: chlorpyrifos in ground black pepper "from India, via Spain". It lists India, Spain and Vietnam as origins, and counts once for each.

Right reading: don't add origin totals together to get an EU total, and read the subject line before blaming a transit country.

Mistake 7: reading the years before 2020 as a trend

This site reads the European Commission's public RASFF Window export. That export effectively starts in January 2020. It also returns 315 older notifications, 283 of them from 2019. The site header's total counts from 23 January 2020 and leaves them out, but they still show up in per-year tables. So the Turkey page lists 1 border rejection in 2018, 11 in 2019 and 289 in 2020. That is not a 26-fold jump in 2020. It is where coverage starts. The Salmonella page shows the same pattern: 41 in 2019, 742 in 2020.

Right reading: start any trend in 2020, and treat earlier years as fragments.

How to cite this site

The data is licensed CC-BY-4.0. Cite it as: Artem Sviridov, "RASFF Trends — EU Border Rejection Notifications", fruit-risk.com, [date you accessed it], https://rasff.fruit-risk.com/. Link the specific page you quote, and name the classification and period it covers. The underlying source is the European Commission's RASFF Window.

Follow new notifications

For a steady view of what is being notified, the weekly pages list every notification with its classification, origin and notifying country. To get them without checking, set up a watch below: one email a week, sent only when there are new notifications.