ATEX harmonised standards: which edition is valid today, and until when

If you run an ATEX zone on site — a spray booth, a powder silo, a solvent transfer skid, a digester — the technical file for that equipment carries the number of a harmonised standard. That number has a date on which it stops conferring presumption of conformity. The date is already settled. And it is written neither on the standard you paid for nor in the official list you copied it from. One of those dates falls five days from now.

The list has no column for dates

The list is Commission Implementing Decision (EU) 2022/1668 of 28 September 2022. It gives effect to Article 12(1) of Directive 2014/34/EU: “Products which are in conformity with harmonised standards or parts thereof the references of which have been published in the Official Journal of the European Union shall be presumed to be in conformity with the essential health and safety requirements set out in Annex II covered by those standards or parts thereof.” Outside that publication the presumption does not exist.

Annex I is the list proper, and it carries two columns only: “No” and “Reference of the standard”. No date, on any row.

A column of dates exists in Annex II, headed “Date of withdrawal”. It holds nine rows, with dates between 27 November 2022 and 29 March 2024. Article 3 says what it was for: Communication 2018/C 371/01 is repealed, but “it shall continue to apply in respect of the references of harmonised standards listed in Annex II to this Decision until the dates of withdrawal of those references”. Since March 2024 that annex has recorded nothing.

Ten withdrawals decided, none recorded

Meanwhile the act has taken on seven amendments, one corrigendum and nine consolidated versions. Every amending decision makes the same two moves in its own annex: it deletes the old row — “row 82 is deleted;” — and inserts a new one with the updated edition. The day the deletion bites is not there: it sits in Article 2, in a single line. In the latest amendment it reads: “Point (1) of the Annex to this Decision shall apply from 4 March 2028.”

Counted on the primary source, ten dates are already settled:

row 82, EN 60079-29-1:2016 — 17 September 2024, Decision (EU) 2023/601;
row 60, EN 50104:2010 — 2 February 2025, Decision (EU) 2023/1587;
row 42, EN 14983:2007 — 1 October 2026, Decision (EU) 2025/597;
rows 10 (EN 1953:2013), 12 (EN 12621:2006+A1:2010), 28 (EN 14373:2021) and 61 (EN 50176:2009) — 12 March 2027, Decision (EU) 2025/1810, whose annex provides in one stroke that “rows 10, 12, 28 and 61 are deleted;”;
row 22, EN 13852-1:2013 — 14 July 2027, Decision (EU) 2026/82;
row 16, EN 13616-1:2016 — 5 October 2027, Decision (EU) 2026/989;
row 81, EN 60079-28:2015 — 4 March 2028, Decision (EU) 2026/1967.

Of those ten dates, Annex II holds none. The two already past left in Annex I only a long dash where the row was: the consolidated text shows that the reference has gone, not since when.

The text of the EN standards is paywalled and we have not read it: number, edition and status are reported as they appear in the Official Journal.

Today the list still tells you yes

The consolidated version in force as we write is 008, of 5 May 2026: at row 42 it still carries EN 14983:2007, immediately followed by row 42a with EN 14983:2024. An engineering office opening the list this morning reads that the old edition counts. Five days from now version 009 takes over and that row becomes a dash, with no notice sent: the decision ordering it dates from March 2025. We have written before about the consolidated versions of a single act; here the trouble is the date, not the language.

Two standards that change without changing number

The nastiest trap sits in the rows where the string stays identical. Row 82 carried EN 60079-29-1:2016; row 82a that replaced it carries the same EN 60079-29-1:2016, plus EN 60079-29-1:2016/A1:2022 and EN 60079-29-1:2016/A11:2022. Anyone matching the string sees no difference — and since 17 September 2024 the edition without those two amendments covers nothing.

Same shape at row 28: EN 14373:2021 leaves on 12 March 2027, replaced by EN 14373:2021+A1:2025. The corrigendum of 3 March 2026 also came in here: the original annex to Decision 2025/1810 read EN 14373:2021/A1:2025, the consolidated text carries EN 14373:2021+A1:2025. One character changes and the object changes: the amendment document alone, or the standard that includes it.

Eighteen months, except once

The pattern looks like a rule: from publication in the Official Journal to withdrawal, exactly eighteen months. It holds for six of the seven. The seventh does not: Decision 2026/989, published on 5 May 2026, sets withdrawal at 5 October 2027, seventeen months. The recital justifying the deferral is identical in all of them and carries no figure — “In order to give manufacturers sufficient time to prepare for application of harmonised standard EN IEC 60079-28:2026, it is necessary to defer the withdrawal of the reference of harmonised standard EN 60079-28:2015” — so the span cannot be inferred: it has to be read in Article 2 of each decision.

What the law puts on you

The Directive does not leave you watching. Article 6(4) is explicit: “Changes in a product design or characteristics and changes in the harmonised standards or in other technical specifications by reference to which conformity of a product is declared shall be adequately taken into account.” Annex III wants in the file “a list of the harmonised standards applied in full or in part the references of which have been published in the Official Journal of the European Union”, and Annex X repeats it at point 6 of the EU declaration. Article 6(3) imposes ten years of retention: ten years in which that document must stay readable against the list in force on the day it was signed, not against today’s.

How we close it

On the complying axis: a control that runs over your technical files, not an opinion. Every ATEX reference pulled from the documents and tied to the Annex I row holding it, with the withdrawal date taken from Article 2 of the decision that ordered it; matching on the full string, A1 and A11 amendments included, because that is where the eye fails; an alert on every new amendment, and a dated trail to put in front of the notified body. On the deciding axis: the same setup holds in one operating model the technical files, the register of equipment installed and sold, the zone classification, the certificates and the monitoring of the Official Journal — and on that model AI agents run the repetitive checks, with your engineer in command. It holds for manufacturing as for energy, defence, public administration and healthcare. We do it on-premise, on standalone machines that need no deep integration into your network, or on our dedicated cloud, with data centres in Italy and premises we staff ourselves, always with shared management: you need no one in-house administering AI models.

How many of your ATEX files cite a reference whose withdrawal date is already set? Half a day with one of our engineers tells you row by row.

Sources

What we do not know

We read the ten dates in Article 2 of the seven decisions and in the consolidated text at 1 October 2026: we do not know whether a later amendment shifts any, because a decision enters into force on the day of publication, with no warning. Of the 3 March 2026 corrigendum we have seen the effect, not the text: the identifier we tried does not return it. We do not know why one decision out of seven grants seventeen months instead of eighteen: the recital does not say. The pace has gone from two amendments a year to three in 2026, but we do not know how many rows will be revised: a standardisation programme is not a calendar. And we do not know how a notified body will treat a file citing EN 60079-29-1:2016 without the two 2022 amendments: the row still exists under that number, and the case has no public precedent.