How this is counted

Source

EFSA publishes every pesticide residue result reported by national authorities, one row per sample and substance, as yearly files per reporting country on Zenodo (CC BY 4.0). This site uses the files for 2020–2024 listed in the table below. The UK layer comes from Defra's Pesticide Residues in Food programme (Open Government Licence). Border-check frequencies come from Commission Implementing Regulation (EU) 2019/1793, every consolidated version since 2020.

One sample, one series

A sample is counted once, in exactly one series, by why it was taken. EFSA's own codes are the starting point, not the last word:

1. EU border check: reported under the EU increased-control programme for imported food; or taken at a border post or import point, from outside the EU, for a product and origin listed in Reg. 2019/1793 for pesticide residues on the sampling date. The second rule exists because some years file border checks under the general programme as "objective" sampling (the Netherlands in 2022: 1,167 recovered checks).
2. Random market sample: objective or simple random sampling, and not a border check.
3. Risk-based sample: selective or risk-based sampling. Never shown as a rate, because inspectors chose these where they expected problems.
4. Follow-up sample: suspect sampling after an earlier finding. Never shown as a rate.

Samples with an unknown origin, and products outside the EU pesticide product list (processed foods) or from animals (except honey), are excluded.

Over the limit

A sample is over the limit when the reporting lab evaluated at least one result as above the maximum residue level. Results above the limit but compliant once measurement uncertainty is allowed for are counted separately and never added in.

When a rate is shown

• Border checks: per product, origin and year, from 30 checks. Never pooled across years: the regulation, the check frequency and exporters all change between years.
• Random samples: pooled across reporting countries and years, from 30 samples, always with a 95% Wilson interval.
• No rate is ever averaged across products or origins.

Reg. 2019/1793 versions

Each sample is judged against the version in force on the first day of its sampling month. Versions used: 2020-05-27, 2020-10-26, 2021-05-05, 2021-11-23, 2022-01-06, 2022-07-03, 2023-02-16, 2023-06-27, 2024-02-06, 2024-07-02, 2025-01-08, 2025-08-12, 2026-02-18, 2026-02-26, 2026-06-30. The original 2019 act, in force until 27 May 2020, uses a different layout and is not parsed; samples from January to May 2020 are judged against the May 2020 version. In two consolidations (June 2023, February 2024) EUR-Lex writes each country once per block; their rows are assigned using the neighbouring versions.

Files on hand

Samples per reporting country and year; border checks split by how they were identified; share with unknown origin. A reporter-year with few border checks may have filed them under another code (Belgium 2021).

Reporting countries not in the table yet are being added; pages appear as soon as a pair reaches a floor.