A supplier sends a twenty-page EMC and safety report from a laboratory in Shenzhen. The European buyer's compliance team replies with one line: "We need this from an EU laboratory." The supplier forwards the email and asks whether the report is worthless.
Usually it is not worthless. Occasionally it is worth nothing. The difference is rarely the country the test was performed in.
The short answer
EU law does not require CE testing to be carried out in Europe. Neither the EMC Directive 2014/30/EU nor the Low Voltage Directive 2014/35/EU names a laboratory, an accreditation scheme, or a country in which testing must take place. The manufacturer draws up the technical documentation, performs the conformity assessment, and signs the EU Declaration of Conformity. What decides whether a Chinese report is accepted is four things: whether the standard and edition are the ones currently cited in the Official Journal, whether the laboratory was accredited for that exact standard, whether the report describes the product actually shipped, and whether the document set behind it holds together. There are exactly three situations where a laboratory outside the EU cannot do the job at all, and they are narrower than most suppliers expect.
What the directives actually require - and what they do not
The confusion starts because "CE marking" is treated as one thing. It is a set of directives, and each one defines a conformity assessment route rather than a testing location.
| Question | Position under EU rules |
|---|---|
| Must CE testing be done in an EU laboratory? | No. For EMC and low voltage products, and for the self-declared radio route, the directives set no laboratory location requirement. |
| Must the laboratory be accredited? | Not mandated by the EMC or Low Voltage Directive. It is what buyers, marketplaces and notified bodies expect to see, and it is where most rejections actually come from. |
| Who signs the Declaration of Conformity? | The manufacturer. If you put your own brand on the product, you are the manufacturer for these purposes, even if a factory in China built it. |
| Who can issue an EU-type examination certificate? | Only a notified body designated by an EU Member State. Not a laboratory, and not a certification company outside the EU. |
| Can an authority demand new testing? | Yes. Under Article 21(4) of the Radio Equipment Directive, a market surveillance authority may require testing by a body acceptable to it, at the manufacturer's or importer's expense. |
| Who has to be reachable inside the EU? | For EMC, low voltage and radio equipment, Regulation (EU) 2019/1020 requires an economic operator established in the EU for products covered by those directives. |
In practice, EMC and low voltage compliance is self-declared. Nobody stamps the file. The manufacturer assembles the technical documentation, keeps it for ten years, signs the DoC, and affixes the mark. A laboratory report is evidence behind that declaration, not an approval - a distinction we covered in test report vs certificate.
So the location question is the wrong first question. A report from Bavaria and a report from Guangdong are both evidence; what differs is the weight it carries.
Where "accredited" stops being optional
ISO/IEC 17025 accreditation is granted per test method and per standard, not to a company as a whole. A laboratory can be accredited for EN 55032 emissions and not for the EN 55035 immunity clauses of the same product file, or for EN 55014-1 household appliances but not for EN 300 328 radio. The accreditation body publishes a scope schedule listing every standard, edition and method covered; the report should carry that accreditation number and symbol on its front page.
No EU directive forces you to use an accredited laboratory. But three practical doors close without one:
- Buyer document review. Large importers and retail compliance teams routinely check that the issuing laboratory's scope covers the standard cited on the report. When it does not, the report is set aside regardless of the measurements inside it.
- Marketplace checks. Amazon's European compliance requests typically ask for reports from an ISO/IEC 17025 accredited laboratory. The seller, not the factory, receives the removal notice.
- Notified body files. Where a radio product has to go through EU-type examination, the NB examines the technical documentation and the evidence behind it. Unaccredited data is not automatically refused, but it invites questions that cost weeks.
A CNAS-accredited report from China sits inside the ILAC mutual recognition arrangement, the mechanism that makes it readable to a European reviewer. That is a factual point worth putting in front of a buyer who says "EU lab only" - before agreeing to retest.
Why Chinese reports get rejected - the six real reasons
In the rejections we see, the country of the laboratory is almost never the stated cause. These are.
1. The standard edition is outdated
Products change, standards get amended, and the Official Journal citation moves. A report citing an edition that has since been superseded still describes a real measurement, but it no longer carries a presumption of conformity against the current harmonised standard. This is the single most common technical reason a European reviewer pushes back, and it has nothing to do with geography.
2. The scope does not cover the standard on the report
Accreditation marks get borrowed. We have seen reports where the laboratory held accreditation for the household appliance emission standard but the product was a multimedia device, or where the logo on the cover belonged to the parent company's calibration scope rather than the testing entity that ran the job. Checking the scope line takes five minutes and is described in how to choose a testing laboratory.
3. The report covers a different configuration than the one shipped
The report names a model, an adapter, an antenna and a set of radio modes. The carton contains a later board revision, a second-source adapter, or an extra radio mode enabled in firmware. On paper the report is fine; against the shipment it is a coverage gap. Importers check exactly this, because it is the gap that turns into a recall.
4. There is no measurement data behind the conclusion
A one-page document stating "the product complies with EN 55032 Class B", with no plots, setup photographs, instrument list or test dates, is not a test report. It is a statement, a consultancy's certificate of conformity, or a template - none of which a technical file can stand on.
5. It is a module report, presented as a finished product report
A radio module with its own approval does not automatically approve the product built around it. The host's DC-DC converter, antenna placement and enclosure change the emission profile, so buyers ask for the host-level assessment while the supplier only holds the module grant.
6. The document set is incomplete
The report is real and the standard is current, but there is no DoC, or the DoC names a company that is not the brand on the label, or the technical file exists only as a folder on a factory computer in Chinese. That is a documentation failure, not a testing failure, and it is the cheapest one to prevent.
The three cases where a Chinese report genuinely cannot work
This is where we tell clients to stop arguing with the buyer.
Case 1: the radio route requires a notified body
Under Article 17(4) of the Radio Equipment Directive, where harmonised standards for spectrum use (Article 3(2)) or an activated Article 3(3) requirement are not applied, are applied only in part, or do not exist, internal production control is no longer available. The product must go through EU-type examination (Annex III) or full quality assurance (Annex IV). Both are notified body procedures, and a notified body must be established under the national law of a Member State. A laboratory in China cannot issue that certificate, and no amount of local accreditation changes it.
This bites most often on proprietary sub-GHz modes with no dedicated harmonised standard, on products where only part of a standard is applied, and on connected products where the cybersecurity requirements activated under Article 3(3) are not fully covered by the cited standard. The route logic is set out in when a notified body is needed.
Case 2: the authority names the body
Article 21(4) allows a market surveillance authority to require a test by a body acceptable to it, at the expense of the manufacturer or importer, when the technical documentation does not present sufficient evidence. "Acceptable to the authority" is the operative phrase: the question becomes who the authority will listen to, not what your report says.
Case 3: the buyer's own vendor approval
This is contractual, not legal, and it is the most common version of the "EU lab" email. Some retail groups, automotive suppliers and medical customers approve only laboratories on their own list, or require reports accredited by a European accreditation body. You can explain the law politely and still have to test where the contract says. Ask for the requirement in writing before quoting a retest, so you know whether it is a policy preference or a hard clause.
Other markets have their own designated-body structures - Korea's KC scheme, India's BIS registration and Japan's technical conformity mark all run through locally recognised bodies - so a Chinese report is an input rather than the approval there as well.
How to check a Chinese report before paying for a retest
Run these checks against the PDF. Most rejected reports fail at least one, and most of those are fixable at the source rather than by starting again in Europe. If the answer is a full repeat, budget for sample shipping, duty and a schedule reset - not just the bench fee.
- Standard and edition on the report against the version currently cited for that directive. If the edition moved, ask the laboratory whether a delta assessment is possible.
- Accreditation scope - look up the laboratory in the accreditation body's public register and search the exact standard number.
- Accreditation symbol and number on the report itself, not only in an email signature.
- The tested configuration: model, hardware revision, firmware version, adapter model, antenna, cable lengths, radio modes. Compare it with the packing list.
- Raw data and setup photographs. Radiated and conducted plots, immunity test levels, photographs showing how the sample was arranged.
- Instrument list, calibration status, and the dates of sample receipt and testing.
- Deviations and exclusions, if any, plus the rest of the file: a DoC naming the correct legal manufacturer, technical documentation in a language the importer can read, and an EU economic operator who will answer an authority's email.
Where we fit
SZC-LAB tests in China and issues reports under ISO/IEC 17025 accreditation, and we are straightforward about what that does and does not buy you. We will tell you when a report we issued is still usable for a European buyer and when it is not, including the cases where the honest answer is that the radio route needs a notified body we are not. What we can do is make the report defensible: the exact standard edition, the accreditation scope line, the tested configuration with adapter and antenna part numbers, the firmware version, setup photographs and the full data set - the six things a European reviewer actually opens. If a buyer has rejected a report, send it to us with their written objection and we will mark up what is genuinely deficient and what is simply unfamiliar to them. Where a project needs a body we are not - an EU notified body, or a scheme that only accepts locally designated laboratories - we say so at quotation stage rather than after the deposit, and hand the data over so the same measurements are not paid for twice. Our overall scope is on the testing and certification page.
The paragraph worth remembering
CE marking does not require a European laboratory. For EMC and low voltage products the directives require a technical file and a signed Declaration of Conformity, not a test performed in a particular country, and a CNAS-accredited report from China is evidence a European importer can legally rely on. What makes a report unusable is narrower: a superseded standard edition, an accreditation scope that does not include the standard cited, a tested configuration that does not match the shipment, missing measurement data, or a radio route that Article 17(4) of the Radio Equipment Directive sends to a notified body - which must be established under the law of an EU Member State. Check those five before you agree to retest everything.
If a buyer has rejected your test report, send us the report and their objection in the original wording. We will go through it line by line and tell you which findings are real, which can be closed with a delta test, and which mean the radio route needs a notified body. No retest quote until that review is done.