ISO/IEC 17025: What an Accredited Calibration Certificate Actually Certifies

What the standard is for

ISO/IEC 17025:2017 sets out general requirements for the competence, impartiality and consistent operation of laboratories. It applies to testing and calibration laboratories alike, and it is used by accreditation bodies as the basis for assessing them.

It is not a product standard and it does not certify an instrument. A brake tester is not "17025 certified". The laboratory that calibrated it may be accredited to ISO/IEC 17025 for the relevant quantity, and those are different statements that are routinely conflated in sales literature.

The 2017 edition put more weight on impartiality, on the reporting of decision rules, and on risk-based thinking in the management requirements, which is why older certificates and newer ones do not necessarily carry the same information.

Three documents that are not the same thing

A manufacturer certificate, an accredited calibration certificate and an accreditation certificate are three different objects, and a procurement file often contains only the wrong one.

The useful question is not "is it calibrated" but "who calibrated it, for which quantity and range, and with what uncertainty". The answers live in the certificate and in the laboratory's accreditation scope, and both have to be checked.

What has to be on the certificate

A certificate that satisfies an assessor is specific. It identifies the instrument, the reference standards used and their own traceability, the conditions, the results, and the uncertainty attached to them.

Vague certificates are not usually fraudulent; they are simply produced outside an accredited system, where the pressure to state uncertainty and a decision rule does not exist.

• Identification of the instrument, including serial number and the range calibrated.
• The reference standards used, with their own certificate references and validity.
• Environmental conditions where they affect the result.
• Indicated value, reference value, error and maximum permissible error at each test point.
• The expanded measurement uncertainty, with the coverage factor and confidence level.
• The decision rule applied, where a statement of conformity is made.
• Date of calibration, date of issue and the next due date.

Uncertainty, the decision rule, and why two labs disagree

Every measurement carries uncertainty, and the standard requires it to be stated rather than implied. When a result sits close to the maximum permissible error, the uncertainty decides whether it can honestly be declared conforming.

A simple acceptance rule ignores uncertainty and compares the error directly with the limit. A guard-banded rule shrinks the acceptance zone by some fraction of the uncertainty, so that a result is only declared conforming when it is conforming with confidence. Two laboratories can therefore reach different conclusions from the same instrument without either being wrong — they applied different rules.

This is why the decision rule has to be written down in the certificate and, better, agreed before the calibration. A station that discovers the rule at an audit is a station that may have to recalibrate and retest.

What this means when you buy a brake tester

The instrument and the calibration regime are one purchase decision. A tester whose calibration coefficients cannot be locked, or whose adjustment history is not recorded, cannot be kept inside an accredited regime without extra paperwork at every calibration.

Traceability also has to survive between calibrations. That is what a period check is for: an internal check against a known load or a reference vehicle, run often enough that drift is caught before the next audit rather than at it.

If your market requires accredited calibration, ask the supplier who performs it locally before you buy, not after. The instrument is portable; the accreditation is not.

• Calibration access in the operator menu, with logged changes and a locked state.
• A period-check procedure the station can run itself, supplied with the machine.
• Exportable raw data, so a recalculation is possible without the manufacturer.
• A named local laboratory accredited for force and, where relevant, for mass.