EA-4/22:2025 — EA guidance on accrediting pesticide residue analysis in food and feed: validation under SANTE/11312/2021, a public list of tests for flexible scopes and less frequent proficiency testing for many commodity groups
On 25 September 2025 European co-operation for Accreditation (EA) endorsed a new edition of EA-4/22 G:2025, applicable from publication with no transition period. It harmonises the criteria accreditation bodies use to assess laboratories testing pesticide residues — methods, validation, quality control, contract review, reports and flexible scope.
- Technical criteria follow the European Commission’s SANTE/11312/2021 document; standardised methods and EU reference laboratory methods are preferred, and mass spectrometry is used for confirmation. Each pesticide’s LOQ should be below the legal limit in the matrices analysed or below a lower limit set by the customer; for complex residue definitions SANCO/12574/2014 applies.
- Quality control has three levels: in every batch at least a recovery at the LOQ or reporting limit, a surrogate in each sample, checks of instrument response and the calibration curve; periodically blind and duplicate samples and reference materials; and proficiency testing — annually during surveillance. With more than three food or feed groups, the participation cycle is five years (two years in the 2018 edition).
- For a flexible scope the laboratory maintains a public list of tests (PLT) with product families and groups, validated matrices, pesticides, LOQ or LOD and the method. A new matrix in a group must be verified at least by recovery at the LOQ, and a new pesticide must be validated. Widespread issuing of results within the scope without the accreditation mark is a serious breach.
- The new edition refers to organic production (Regulations (EU) 2018/848 and 2021/1165), where mere detection of residues matters. SANTE/11312/2021 now exists as version v2026, applicable from 1 January 2026: it adds, among other things, a sub-sampling variability criterion, requirements for background levels in blanks and an uncertainty example for complex residue definitions; it keeps 70–120% recovery with RSDr ≤ 20% and a default expanded uncertainty of 50%.