One in Three Fails: What EU Enforcement Data Reveals About Chemical Documentation

DH Scientific — the reference implementation of the MOL-GOD platform.
Every statistic in this article links to an official report or press release of the European Chemicals Agency (ECHA) or to EU legal texts on EUR-Lex. Methodological limitations of the data are stated explicitly at the end. We publish nothing we cannot source — because that is literally what our platform does.

Your product is good. Your price is competitive. Your GHS Safety Data Sheet looks professional — 16 sections, pictograms, hazard statements. And yet your EU importer’s compliance officer rejects it, or worse: the shipment is stopped at the border.

This is not an anecdote and it is not scare marketing. It is what EU-wide enforcement data shows. The EU runs coordinated inspection campaigns — the REACH-EN-FORCE (REF) projects of ECHA’s Enforcement Forum — in which national inspectors across ~29 EEA countries check the same obligations using a common methodology, and publish the results. We read those reports so you don’t have to. Here is what the three most relevant campaigns found — and then, precisely, what MOL-GOD does about each finding.

The official numbers

REF-11 (2023): 35% of inspected Safety Data Sheets were non-compliant

In ECHA’s most recent SDS-focused enforcement project, inspectors in 28 EU/EEA countries checked over 2,500 Safety Data Sheets against the current format (Regulation (EU) 2020/878). The results, published by ECHA on 3 December 2024:

Finding	Rate	Source
SDS non-compliant overall	35%	ECHA press release, 3 Dec 2024
Missing nanoform information (where required)	67%	same
Missing endocrine-disruptor information (where required)	48%	same
Data-quality deficiencies (hazard identification, composition, exposure controls)	27%	same
Missing required exposure scenarios	18%	same
Missing information required by authorisation decisions	16%	same

Full report: Forum REF-11 project report on Safety Data Sheets (PDF, ECHA, Dec 2024).

Enforcement was real, not theoretical: ECHA reports that inspectors issued written advice, administrative orders and fines, and in some cases filed criminal complaints.

Two things make this number remarkable. First, these were overwhelmingly EU-based suppliers operating inside the EU regulatory system — and still more than one in three documents failed. Second, the long-term trend, discussed in ECHA’s own Safer Chemicals podcast on REF-11, shows how persistent the problem is: comparable checks found roughly 48% non-compliance about two decades ago (CLEEN network project) and 52% in the REF-2 project (2013). After thirty years of the SDS system, a third of documents still fail.

The conclusion that matters commercially: producing a compliant SDS is not a translation task and not a template task. It is a data problem — and one that even native EU suppliers get wrong at scale. That data problem is exactly what MOL-GOD was engineered to solve.

REF-12 (2024): imports checked at the border — 1 in 3 substances in mixtures had no REACH registration

The most directly relevant campaign for non-EU exporters. Inspectors in 29 EEA countries, working with customs authorities, performed 2,603 targeted controls on imported substances, mixtures and articles. Published by ECHA on 10 December 2025:

Finding	Rate	Source
Substances in imported mixtures missing required REACH registration	~1 in 3 (32%)	ECHA press release, 10 Dec 2025
Substances imported on their own missing registration	7%	same
Non-compliance with REACH restrictions among checked products	16%	same
SVHC authorisation missing or expired	4 of 21 checks	same

Full report: Forum REF-12 project report (PDF, ECHA, 2025) — and one detail from that report matters especially for our readers: the largest share of controlled products originated from China (1,334 cases). Chinese-origin shipments are, factually, the primary target group of EU import enforcement.

The consequence ECHA describes is concrete: almost all non-compliant products detected before release for free circulation were either refused entry to the EEA market or admitted only after corrective measures. This is not a fine you pay later. It is your container standing at the border while demurrage accrues, your delivery deadline passes, and your EU customer calls your competitor.

REF-8 (2020): online sales — 78% of checked products breached REACH restrictions

If your route to the EU market is e-commerce, the numbers are the harshest of all. Inspections of nearly 6,000 products sold online, in 29 countries (ECHA press release, 8 Dec 2021; full REF-8 report PDF):

• 78% of products checked against REACH restrictions did not comply.
• 75% of offers breached the advertising obligations of CLP Article 48 (hazard information missing from the online offer).
• The campaign triggered more than 5,000 enforcement actions.
• Non-compliance was higher on marketplaces than in dedicated web shops.

Online sales are being re-inspected right now: the follow-up project REF-13 targeting online sales of chemical products ran its inspections in 2025, with the report expected in 2026 (ECHA Enforcement Forum working groups). Import controls likewise remain a stated priority in the Forum’s 2026–2027 work programme. The direction of travel is unambiguous: more checks on imports, more checks online. The window for „nobody will look at my documentation” is closing, measurably.

Why non-EU Safety Data Sheets fail: the seven structural gaps

The enforcement data above measures outcomes. The causes, for documents originating outside the EU system, are structural — an SDS written for China GHS or US HazCom is missing entire categories of legally required content. Each item below links to the legal basis so you can verify it:

1. Language. REACH Article 31(5): the SDS must be supplied in an official language of the Member State where the product is placed on the market (REACH consolidated text). English alone is sufficient only for Ireland and Malta. Poland requires Polish; Germany requires German. This failure is visible in one second.
2. EUH statements. The EU applies supplemental hazard statements (EUH014, EUH019, EUH031, EUH066, EUH208…) defined in the CLP Regulation (consolidated text). These codes do not exist in GB 30000 or in US HazCom — no Chinese or American template will ever generate them.
3. DNEL and PNEC values. EU SDSs for registered substances must state Derived No-Effect Levels and Predicted No-Effect Concentrations — outputs of the REACH registration process with no equivalent in other GHS implementations.
4. National occupational exposure limits in Section 8. The SDS must cite the destination country’s national limits (Polish NDS, German AGW, French VLEP…), not only EU indicative values — and they genuinely differ. Verifiable example: for acetone, the EU indicative limit is 1210 mg/m³ (Directive (EU) 2019/1831 / IOELV framework), while the binding Polish national limit (NDS) is 600 mg/m³ (Polish regulation on NDS/NDN, Dz.U. 2018 poz. 1286, as amended) — twice as strict. An SDS showing only the EU value gives a Polish employer legally wrong data for workplace risk assessment. A compliant SDS is therefore country-specific, not merely language-specific.
5. Classification from the wrong source. Where a substance has a harmonised classification in CLP Annex VI, that classification is legally binding — verifiable per substance in the official ECHA C&L Inventory. Two common failures: self-classifying more leniently than the mandatory Annex VI entry (illegal), or copying the union of all public self-classification notifications (inflating the SDS with phantom hazards that raise your customer’s transport and storage costs and collapse trust on audit). Note that classification and labelling errors in Sections 2.1/2.2 were among the concrete deficiency types inspectors recorded in REF-11.
6. Section 15 without EU content. REACH restriction status (Annex XVII list), authorisation status (Annex XIV list), SVHC Candidate List status (official list) and applicable national law must appear. A Section 15 citing Chinese regulations is an immediate marker of a non-EU document — and it hides exactly what an EU importer legally needs to know before purchase.
7. UFI and Poison Centre Notification for hazardous mixtures. Annex VIII to CLP requires notification via the ECHA Poison Centres portal and a UFI code on the label and in SDS subsection 1.1. REF-11 inspectors specifically recorded missing or misplaced UFI codes as a recurring deficiency. Without the exporter’s cooperation on composition disclosure, the EU importer cannot generate a UFI at all — and the product cannot legally be placed on the market.

The cost asymmetry every exporter should calculate

Put the two sides on one sheet of paper.

The cost of documentation failure is not the cost of reprinting a PDF. Per the REF-12 findings, non-compliant goods detected at the border were refused entry or held pending corrective measures. In commercial terms that means: demurrage and storage while the container waits, potential return freight or destruction of goods, contractual penalties for missed delivery, and — hardest to recover — a compliance officer at your EU customer who now marks your company as a supplier risk. In a B2B chemical relationship, that flag rarely gets removed. You do not lose one shipment; you lose the account.

The cost of getting documentation right is a fraction of a single held container — and unlike demurrage, it is an investment that compounds: the same verified data layer serves every product, every language, every shipment, every audit.

That asymmetry is the business case for MOL-GOD in one sentence: it is dramatically cheaper to be auditable before the border than to become compliant after it.

Why MOL-GOD exists — and what it actually does

MOL-GOD (molgod.org) was built for one purpose: to raise the compliance quality of chemical product documentation and reduce the risk of import rejection caused by formal deficiencies. Not as a slogan — as an architecture.

We built it the hard way: by auditing our own live catalog on this very site and finding the same classes of failure that ECHA’s inspectors find across Europe — misclassified hazards, missing supplemental statements, wrong exposure limits, unverifiable references. Instead of patching documents one by one, we rebuilt the data layer so that those failures become structurally impossible to publish. The site you are reading runs on the result.

The core rule of the platform mirrors the standard we applied to the statistics in this article: every affirmation on a product page or SDS must trace to a verified authoritative source — or be explicitly marked as unverified. No source, no claim.

Here is how that maps, point by point, onto the failure modes EU inspectors actually record:

What inspectors find (REF-11/REF-12)	What MOL-GOD does	Your risk reduced
Incorrect classification & labelling (Sec. 2.1/2.2)	Classification gated on CLP Annex VI harmonised entries first; self-classification data displayed separately, clearly labelled, never mixed in	No illegally softened hazards; no phantom hazards inflating your customer’s transport and storage costs
Supplemental statements missing	EUH statements are native to the phrase canon — not an afterthought a filter can silently drop	Section 2 complete on the first pass, not after the third rejection
Wrong or missing exposure data (Sec. 8)	Per-country exposure-limit canons (e.g. the full Polish NDS table sourced from the national regulation) — never a single hardcoded EU value	The Polish, German or French employer reading your SDS gets legally correct workplace data
Outdated or unverifiable content	Three-state provenance on every field: VERIFIED / AGGREGATE / UNKNOWN, source one click away	When the compliance officer audits your document, the audit trail already exists — you pass in minutes, not weeks
Language deficiencies	Language versions built from the official CLP phrase libraries — the EU publishes every H/P/EUH statement in all 24 official languages, so rendering is deterministic lookup, not machine translation	Adding a new EU market is a data operation, not a translation project
Missing UFI / PCN elements	UFI and notification fields are first-class citizens of the document model	Your importer can actually fulfil the Annex VIII duty — which makes you the supplier they can work with

Notice what is absent from that table: artificial intelligence writing safety content. MOL-GOD deliberately does not generate hazard data with a language model. Regulatory phrases are deterministic lookups from official EU sources; numerical data carries provenance; anything unverified says so, visibly. In a domain where a fabricated number can hurt a human being handling a chemical, „the AI wrote it” is not an acceptable answer — so we removed the possibility by design.

What MOL-GOD is not — stated plainly

We deliberately do not sell „REACH compliance” as a checkbox. Compliance is a legal state that depends on registration status, tonnage and your role in the supply chain (importer, Only Representative, downstream user) — anyone selling it as a software feature is misleading you. What MOL-GOD delivers is audit-ready, provenance-backed documentation: the data layer that makes compliance achievable, demonstrable, and cheap to maintain.

Auditable, not merely „compliant.” We chose that phrase because an audit is the moment your documentation is actually tested — by an inspector, by customs, by your customer. MOL-GOD is built for that moment.

What this means, honestly stated

We will not tell you that „90% of Chinese SDS get rejected” — no such statistic exists in any official source, and we do not publish invented numbers. What the official record does support is this:

1. Even inside the EU system, 35% of inspected SDSs fail (REF-11) — the document is genuinely hard to get right.
2. Chinese-origin products were the largest group in EU import controls (1,334 cases in REF-12), and about one in three substances in imported mixtures lacked required registration, with non-compliant goods stopped at the border.
3. Enforcement intensity is increasing, with imports and online sales as declared priorities for 2026–2027, and the REF-13 online-sales report due in 2026.
4. A China GHS or US HazCom SDS is structurally incapable of satisfying seven categories of EU requirements without being rebuilt on EU data — this follows directly from the legal texts linked above, independent of any statistic.
5. The economics favour prevention: verified documentation costs a fraction of one border hold — and it is reusable across your whole catalog.

Sources and methodology — read this before quoting the numbers

We hold our own marketing to the standard we hold chemical data to, so three caveats:

1. Enforcement samples are risk-targeted, not representative. ECHA states explicitly in its reports that inspectors often selected products and companies where non-compliance was considered likely, and that results „are not necessarily representative of the situation in the EU-EEA market as a whole” (REF-12 report; the same caveat appears in REF-8). The true market-wide non-compliance rate may be lower than the inspection figures. What the figures do prove: when EU inspectors look, they find failures at scale — and they choose where to look.
2. Different projects measured different obligations. REF-11 checked SDS content quality; REF-12 checked registration, restrictions and authorisation at import; REF-8 checked online offers. The percentages are not directly comparable to each other, and we have not combined them into any aggregate figure.
3. China-origin prevalence in REF-12 reflects trade volume and targeting, not a measured per-country failure rate for chemicals. We state only what the report states: Chinese-origin products were the most numerous group controlled (1,334 cases).

Primary sources used in this article:

Source	Publisher	Link
REF-11 press release (SDS compliance)	ECHA, 3 Dec 2024	echa.europa.eu
REF-11 full project report (PDF)	ECHA	PDF
REF-12 press release (imports)	ECHA, 10 Dec 2025	echa.europa.eu
REF-8 press release (online sales)	ECHA, 8 Dec 2021	echa.europa.eu
Forum enforcement projects overview	ECHA	echa.europa.eu
Forum work programme 2026–2027	ECHA, 24 Nov 2025	echa.europa.eu
REACH (consolidated)	EUR-Lex	eur-lex.europa.eu
CLP (consolidated)	EUR-Lex	eur-lex.europa.eu
SDS format — Reg. (EU) 2020/878	EUR-Lex	eur-lex.europa.eu
CLP Annex VIII (UFI/PCN)	EUR-Lex	eur-lex.europa.eu
Directive (EU) 2019/1831 (IOELVs)	EUR-Lex	eur-lex.europa.eu
Polish NDS regulation (Dz.U. 2018 poz. 1286)	ISAP / Sejm RP	isap.sejm.gov.pl

See the standard in practice

The claims above are testable right now, on this site:

1. Open any product page on DH Scientific and expand the hazard data — every value shows its verification state and its source, one click away. That is not a demo environment; it is our live catalog. Try it on a live product →
2. Read the full legal roadmap: our complete A-to-Z guide for non-EU exporters, with links to every EU legal act: Exporting Chemicals to the European Union — The Complete Legal Roadmap →
3. Run your EU-facing catalog on MOL-GOD. If you export chemicals to Europe and your documentation is the bottleneck, that is exactly the problem this platform was built to remove. Talk to us →

Disclaimer: This article is general information, not legal advice. Verify all legal texts against the consolidated versions on EUR-Lex and consult a REACH compliance professional for your specific situation. Enforcement statistics cited here come from targeted inspection campaigns and should not be read as market-representative rates — see the methodology section above.

DH Scientific is operated by Nonsensia Ltd (Company No. 15165427, London). Science first. Commerce as consequence.