EU border check levels track rejection rates, product by product

The EU's border check levels line up with what its border rejects. Under Regulation (EU) 2019/1793, origin-and-product lines checked at 50% show a median 718 border rejections per $1 billion of EU imports over three years. Lines checked at 10% show 62. Across 22 lines the rank correlation between check level and rejection rate is 0.76.

This is for exporters reading their own check level, and for authorities and journalists asking whether the list points at the right places. We tested the opposite claim, that check levels barely track outcomes. It does not hold.

Check level against rejection rate

Regulation 2019/1793 sets, for each listed origin and product, the share of consignments that get identity and physical checks at the EU border: 10%, 20%, 30% or 50%. We matched each level to the site's border-rejection rate for the same origin and product.

The median rises at every step. The ranges overlap, so a check level is a guide to a product's record, not a forecast.

The pairs that show it

• Turkish pistachios and hazelnuts. Pistachios are under a 50% aflatoxin check and run at 244.0. Hazelnuts run at 1.8, 136 times lower, and they are not on either annex. The 50% line covers pistachios only (CN 0802 51 and 0802 52). Hazelnuts share the four-digit heading, so the two are easy to confuse.
• Peppers. Turkey (20%) runs at 152.5. Egypt (30%) runs at 945.1. Thailand and Vietnam (50%) run at 4,214 and 1,677.
• Sesame. Turkey and India (30%) run at 226.6 and 201.7. Nigeria (50%) runs at 718.4.

Where the pattern breaks

• Argentine groundnuts: 4.7 under a 20% check. US groundnuts sit at the same 20% and run at 199.0. Argentina's line is the clearest case of a check level above what the record now shows.
• Chinese rice: 371.9 and not listed. 15 of its 21 border rejections since October 2023 are for unauthorised genetically modified rice. That hazard has its own emergency measure, Implementing Decision 2011/884/EU, which is still in force. So "not listed in 2019/1793" does not mean "not checked".
• Mixed product groups. Some site pages group a listed product with unlisted ones: vine leaves with grapes, passion fruit with bananas, tomatoes with other vegetables. The trade figure then covers far more than the checked product, and the rate is diluted. We left these 13 lines out of the table. With them included, the correlation falls to 0.62, but the median still rises from the 10% band to the 50% band.

Why the pattern is partly built in

Two things push in the same direction, and this data cannot separate them.

1. More checks find more. A consignment that is never sampled is rarely rejected. A high rate under a 50% check partly reflects the testing itself.
2. The list follows the record. Article 12 of the regulation requires the Commission to review the annexes at least every six months, "to take into account new information related to risks and non-compliance". Lines with many rejections are the ones that get listed or raised.

So the stronger signal runs the other way. A low rate under a high check level is a product being tested often and still passing, and that is the case for lowering the check. Among the lines we could compare, only Argentine groundnuts fit it clearly.

What this means if you…

• …export a listed product: your check level tells you how the EU reads your origin's record. Expect it to move with that record at the next review. If your rate is low for your band, that record is your argument.
• …work for an authority: the list broadly points where the rejections are. Check lines whose rate falls well below their band, such as Argentine groundnuts at 20%, at the next review.
• …write about EU border checks: check the exact tariff line before saying a product is under increased checks. Four-digit codes mix pistachios with hazelnuts, and chilli with sweet peppers.

---

Method. Rates are border-rejection notifications per $1 billion of EU imports of that commodity from that origin, over the 12-quarter window shown on each page (FAOSTAT trade, 2024 the latest year). They are quoted from the site's origin-and-product pages, as of 21 September 2026. Pages below five notifications show no rate and are excluded. Check levels come from the consolidated text of Regulation (EU) 2019/1793 dated 30 June 2026, which includes amendments up to Implementing Regulation (EU) 2026/1206 of 9 June 2026. We matched them by full CN code, not by four-digit heading. Where one line carries several checks (for example Turkish citrus: 20% on lemons and oranges, 10% on mandarins), the highest counts. "Clean" lines are those where the listed product is essentially the whole product group the rate measures: 22 of 35 lines. The correlation is Spearman's rank correlation.

Follow Turkey's border rejections, pistachios and figs included, with a weekly alert: