The short version: CMA is about market access in China, CNAS and IAS are about technical competence
When a third-party test report lands on your desk, resist the urge to flip straight to the data pages. Start with the cover: how many accreditation marks are on it, and which one is which. CMA is China's statutory qualification for inspection and testing bodies. If the report is going to be used to demonstrate product conformity to a domestic regulator, and medical device registration is the classic case, that mark is the entry ticket. CNAS is laboratory accreditation granted by the China National Accreditation Service for Conformity Assessment. It is applied for voluntarily, the technical assessment is carried out against the general requirements for the competence of laboratories (ISO/IEC 17025, with the currently effective version governing), and its value lies in recognition among international peers. IAS is a US accreditation body that assesses against the same ISO/IEC 17025 and is well recognised within the North American system.
These three are not tiers of the same ladder, and none of them substitutes for another. They answer three different questions: whether the body has legal standing inside China to issue data with evidentiary weight; whether its technical competence has been assessed against internationally accepted criteria; and whether that evidence of competence counts for anything in the market you are actually going to. The mistake we see most often is manufacturers collapsing those three questions into one, on the theory that more stamps means more safety. What they end up with is a stack of reports carrying every mark you could want, on test items that sit outside the accredited scope.
One premise has to be stated before anything else, or every method below gets misapplied: an accreditation mark only demonstrates that the laboratory has the corresponding technical competence within its accredited scope. It does not constitute a commitment regarding market access in any target market. Whether a report is accepted is decided by the receiving party under its own rules: the regulator for domestic submissions, possibly a notified body on the EU route, the local competent authority in an overseas market, and your customer in a commercial procurement setting. In all of those settings the accreditation mark is supporting evidence, not the conclusion. What follows is the general principle. Which items in an existing report can be reused for a particular product family, and which still have to be run, depends on the standards applicable to that family and on the specific requirements of the target market. That has to be judged family by family and cannot be derived directly from the general rules in this article.
Side by side: nature, basis, and where each one applies
| Comparison | CMA | CNAS | IAS |
|---|---|---|---|
| Nature | Statutory qualification for inspection and testing bodies in China; mandatory | Voluntary laboratory accreditation | Voluntary laboratory accreditation |
| Main technical basis | Qualification assessment criteria, whose technical requirements align closely with ISO/IEC 17025 | ISO/IEC 17025 | ISO/IEC 17025 |
| Question it answers | Can the body issue data with evidentiary weight to Chinese regulators | Has technical competence been assessed against internationally accepted criteria | Has technical competence been assessed against internationally accepted criteria |
| Typical use | Domestic device registration testing, market surveillance sampling, government-commissioned projects | Export trade, review by overseas customers and notified bodies, international recognition | Customer acceptance in North America, buyer audits |
| Boundary of the mark | Covers only the items and standards listed in the annex to the qualification certificate | Covers only the items and standards listed in the accredited scope annex | Covers only the items and standards listed in the accredited scope annex |
| Common misreading | Assuming CMA makes a report universally valid | Assuming CNAS means the target country's regulator accepts it automatically | Assuming it is equivalent to regulatory clearance |
The row on the boundary of the mark is the crux of this whole article. None of the three marks is a pass issued to the laboratory as an organisation. Each is issued to a combination: this laboratory, using this method, measuring this parameter, within this range. It is entirely normal for the same laboratory to have item A inside its scope and item B outside it. That is not a sign of a problem with the laboratory; it only means the accredited scope does not cover the item you need.
The scope annex is the document you actually need to read
Most disputes do not arise from the certificate, they arise from the annex. The face of the certificate carries only the body's name, a number and a validity period. What actually defines what the laboratory can do is the capability scope annex behind it. When you read that annex, there are at least three columns you have to get straight.
The item name. Names in the annex are usually more granular than the shorthand people use day to day. A single item as you think of it may be split into several entries, or only one of those entries may be listed. What you compare against is the specific entry written on the report, not the broad category it belongs to.
The test basis. The same item performed to different standards can follow completely different assessment logic, and what the annex accredits is the item plus standard combination. If the standard shown on the report does not appear on the corresponding row of the annex, that item carries no accredited status even when the item name matches.
The limitations column. This one gets skipped, and it is often the decisive one. Typical limitations include applicability restricted to certain sample types or materials, coverage of only part of the tests in a standard, issuing data without a conformity statement, or a method carrying specific conditions of use. Where the limitation conflicts with your intended use, the report goes back for rework no matter how good the data looks.
Seven checks, in this order
In practice we advise clients to work through a report in the following order. If the first three do not hold up, the data behind them is beside the point.
One: cover marks and certificate numbers. Marks normally carry a number next to them, and that number is your key to looking up the scope on the accreditation body's official website. If there is a mark with no number, or the number does not match the name of the issuing organisation, ask.
Two: whether the tested items fall inside the accredited scope. This step means actively requesting the relevant pages of the capability scope annex and comparing entry by entry against the three columns above. Anything outside the scope carries no accredited status even if it was tested and reported.
Three: the standard numbers and version status in the test basis field. Check that the version is currently effective and that it matches the list of standards you are submitting. When a standard is revised, both the test method and the assessment logic may change, and a submission built on a superseded version tends to come back as a deficiency letter. For specific clauses, test parameters and acceptance limits, the currently effective version of the standard text governs in all cases. Do not work from memory or from what was true on the last project.
Four: sample information. Model and specification, production lot number, sample quantity, date of receipt, description of sample condition, and sample source, meaning whether the samples were submitted by the manufacturer or drawn by the laboratory. Reports used for registration carry requirements on how representative the samples are, and submitted samples and sampled units do not have the same evidentiary standing. Also confirm that the model designation on the report matches the way it is written in your registration unit exactly. One extra suffix or one missing letter is enough for the two to be read as different products.
Five: how the conclusion is worded. Some reports state conformity; others report data without any judgement. Whether a conformity statement is given, how measurement uncertainty is handled when it is, and what decision rule applies, are all things to agree with the laboratory at the order stage. Most arguments about results near a limit trace back to a decision rule that was never agreed in advance.
Six: the subcontracting statement. Subcontracting some items to another body is normal practice in this industry, but it has to be declared in the report, and you need to check that the subcontractor's credentials meet your intended use. Some overseas customers explicitly refuse subcontracted items, and that kind of requirement belongs in the conversation before the order is placed, not after.
Seven: authorised signatory, issue date and tamper-evident features. Authorised signatories are also authorised by field. A signatory whose authorised field does not match the scope of the report is a defect in its own right.
Why "we have CNAS, so it works everywhere" is a misconception
CNAS participates in international mutual recognition arrangements. What that means is that accreditation bodies in different countries recognise each other's accreditation decisions made under ISO/IEC 17025, in other words they accept at a technical level that the other body's assessment is credible. That is a different thing from a target country's regulator automatically accepting the report. Mutual recognition solves the credibility of the assessment; it does not solve acceptance of a submission.
When you get down to a specific market, the logic runs roughly like this. For domestic registration in China, what matters is the body's statutory standing and whether its testing capability covers the items involved. On the EU route, the report is typically reviewed by a notified body or an authorised representative as part of the technical documentation, and what they care about is whether the testing was carried out correctly against the applicable standards and whether there is objective evidence of laboratory competence; the accreditation mark is strong supporting evidence, not an exemption. On the US route, the focus is on the test plan and conformity with the standards themselves, with laboratory accreditation reviewed alongside as evidence of competence. For the requirements of any specific market, the regulations and guidance currently published by that market's competent authority govern. Do not apply another manufacturer's experience to your own file without checking.
The reverse case is just as common. Some overseas buyers write into their purchasing documents that reports must come from a laboratory accredited to ISO/IEC 17025. In that situation CMA alone is not enough, and you still have to check whether the accredited scope annex covers the item in question. The requirements in that standard concerning validity of results, decision rules, and risk-based planning are precisely the parts overseas customers look at when they review a report, and the currently effective version governs the detail.
Work backwards from the use to decide which report you need
| Your use | Confirm first | Easy trap |
|---|---|---|
| Domestic registration submission | The body's domestic standing, and that its testing capability covers every item in the submission | Some items fall outside scope, a second laboratory has to be found, and the schedule slips |
| EU export or overseas tenders | Whether the ISO/IEC 17025 accredited scope covers the items, and whether an English report is available | A Chinese-only report has to be reissued in English, and there is no time left |
| North American customers or buyer audits | Which accreditation system the customer's purchasing documents name | Only one mark was prepared and the customer does not recognise it |
| Internal design verification and exploratory testing | Whether the method is sound and the data traceable | Paying for a formal report to do exploratory work, or the other way round |
| Complaint handling and failure analysis | Whether a conformity statement is needed, and whether samples must be retained | Samples destroyed, retesting impossible |
Exploratory testing deserves a separate word. Running a round of verification testing before the design is frozen, without issuing a formal report, costs far less than retooling and resubmitting after a formal report fails. Plenty of manufacturers try to save one test fee by sending a design they are unsure about straight into registration testing, get held up on a single item, and lose months to rework. For how projects are scheduled and what the lead times look like, start with the testing process and the sample submission requirements.
ISO 13485 and ISO/IEC 17025 are not interchangeable
When auditing a laboratory, quality departments frequently ask it to produce an ISO 13485 certificate. That confuses two different systems. ISO 13485 is the quality management system standard for medical devices; its subject is device manufacturers and related parties, and it governs the chain running from design and development through production, release and traceability. ISO/IEC 17025 sets out the general requirements for the competence of testing and calibration laboratories; its subject is the laboratory, and it governs the chain running from method validation through equipment traceability, personnel competence, validity of results and completeness of reporting.
Third-party testing laboratories generally do not manufacture devices, so they do not run their own operations under ISO 13485. What you are assessing is laboratory competence, and the matching evidence is the accredited scope annex, method validation records, equipment traceability certificates, personnel authorisation records and participation in proficiency testing, not a management system certificate. The mirror image applies to you: if you as a manufacturer perform some testing in house and want to use the data for design verification, that data needs corresponding method and equipment control evidence inside your own ISO 13485 system. Calibration status within its validity period, a statement of method suitability, qualification records for the operators, and traceability of raw data. Miss any one of those and the data will not stand up in review either.
How to check for yourself whether a report is genuine
You do not need specialist tools. Three steps will get you through a first screening.
First, take the accreditation number from the report cover and look the body up on the relevant accreditation body's official website, confirming that its status is normal and the certificate is within its validity period. Second, request the capability scope annex and reconcile it, finding the corresponding row for every item and every test basis on the report; anything you cannot find is out of scope. Third, call the issuing organisation directly and verify the report number and issue date. Reputable laboratories all have a channel for report lookup and authenticity verification.
That third step looks redundant and is anything but. Falsified reports do exist in the market, particularly reports that have changed hands several times through trading intermediaries. If a buyer has doubts about a report, having them contact the issuing laboratory directly is far less work than explaining back and forth. For more of this kind of question, see the FAQ.
Settle these before you send samples and you save a round of rework
The order form is not a formality. We suggest putting the following in writing at the order stage: the intended use of the report, meaning domestic registration, export, customer audit or internal verification, because the use determines which qualification the report is issued under; the language versions required and whether Chinese and English reports are issued together; whether a conformity statement is needed and under what decision rule; the model coverage strategy for the samples; whether subcontracting is acceptable; how many original copies are needed and whether an electronic version is required; and whether samples are returned after testing and how long retained samples are kept.
The samples themselves also need to be prepared according to the intended use. Samples from a single lot that can be traced back to production records hold up far better in review than an assortment pulled together from wherever; samples should carry the same markings, instructions for use and packaging format as the real product, because the test condition for many items depends on the actual delivered form of the product; and destructive items need enough spare samples set aside according to the test plan, because if something anomalous shows up mid-way and there is nothing left to retest, you are starting over.
One more word on model coverage. Reasoning from load paths, material behaviour and structural margin, the usual approach is to select the specification with the more demanding duty and the smaller margin as the representative one for testing, on the logic that if the low-margin specification passes, the higher-margin members of the same family will generally pass under the same failure mode. That is an engineering judgement, not a statistical conclusion, so three things have to go into the technical file: the basis on which the family was grouped, meaning whether structure, materials and process are consistent; the justification for choosing the representative specification; and an analysis of the differences between the untested specifications and the representative one and how those differences affect the test result. A single line saying that a typical specification was selected will draw a question in review essentially every time. How to make that selection for a given product also has to follow what the applicable standard text actually says.
Leave any of this out and it usually means the report gets amended once it has been issued. Once a report is signed, correcting it means going through an amendment or reissue procedure, which is a good deal more trouble than agreeing the details up front.
What we can offer
SUNGO Lab is a third-party medical device testing laboratory accredited by CNAS, CMA and IAS (USA), with laboratories in Shanghai and Hefei, covering performance, safety, electromagnetic compatibility, biological evaluation, packaging and transport testing for medical devices. If you already have a report and are unsure whether it will hold up for your target market, or you are weighing which route to test under, send us the report and your submission plan and we will work through the accredited scope and test basis with the logic above and tell you what additional testing we think is needed. Whether the report is accepted still rests with the rules of the receiving party. Our capabilities and item coverage are set out under testing services. Call +86 132 4819 8029 or request a quote.