ISO/IEC 17025 clause 7.6: explain measurement uncertainty

What does the clause require?

Identify significant sources of uncertainty, including sampling where relevant, and use suitable analysis. A laboratory performing calibrations must evaluate uncertainty for its calibrations, including calibration of its own equipment. A testing laboratory must also evaluate uncertainty, with appropriate estimation where the method prevents a rigorous evaluation.

The standard recognises established methods and circumstances where previously determined uncertainty can remain applicable if critical influences are controlled. This does not mean copying an uncertainty value indefinitely. The laboratory needs to understand the basis and whether the actual method, range and conditions remain within it.

A practical sequence for the laboratory

Sterling recommends beginning with the result being reported and the method producing it. Identify the quantities, equipment, references, preparation and conditions that can influence it. Decide which contributions are significant and how they will be evaluated. The appropriate technical approach depends on the measurement; there is no useful universal budget for every laboratory.

Use evidence appropriate to the contribution, such as relevant calibration information, method performance or repeat measurements. Document the reasoning and limitations so another competent person can understand it. Review the evaluation when method changes or performance evidence suggest the previous basis is no longer suitable.

Connect this work to reporting and conformity decisions. The uncertainty evaluation under 7.6 and the reporting requirements under 7.8 are related but different responsibilities. An internal calculation does not establish that a customer received the information needed to interpret a result.

Why clause 7.6 matters to the laboratory

Uncertainty is how a laboratory tells the customer how far a result can be relied on. Without it, a number looks more exact than it is, and decisions that depend on it, such as accepting a batch or approving a repair, may be taken on false confidence. A properly built evaluation also helps the laboratory itself: it shows which parts of the process contribute most doubt, so improvement effort goes where it matters. It becomes crucial near specification limits, where the margin for a wrong conclusion is small. Laboratories that treat the work as a shared, understood analysis, not a number borrowed from elsewhere, find that questions from customers and assessors become easier to answer.

How to implement clause 7.6

Build the evaluation from the actual measurement process, then keep it alive. Start with one high-volume result and extend from there, so the team learns the approach on something familiar.

1. Choose one reported result and describe the measurement process behind it in plain steps.
2. List the influences at each step, such as the reference, the instrument, preparation, environment, the operator and sampling where relevant.
3. Decide which influences are significant, and what evidence will quantify each: calibration data, repeat measurements, validation results or technical knowledge.
4. Combine them with a method suited to the case and record assumptions, so another competent person can follow the reasoning.
5. State where the evaluation applies: method, range, conditions and the people allowed to use it.
6. Set review triggers, such as a method change, new equipment or monitoring data that no longer match the evaluation.
7. Link the outcome to contract review and reporting so the customer receives uncertainty information that matches the agreed use of the result.

Calibration example: uncertainty beyond the reference certificate

Construction materials example: variation in the actual test

A materials testing laboratory has an established uncertainty evaluation, but a new preparation arrangement changes how specimens are handled. Management wants to keep the old figure to avoid delaying customer work. The technical question is whether the earlier basis still represents the method's performance.

The laboratory evaluates the change and any significant contributions, updates the basis where necessary and checks associated reporting. This fictional scenario shows why uncertainty is connected to operational changes. A value is not reliable merely because the same number appeared on last year's report.

Manufacturing example: a result near a specification limit

A manufacturer asks its laboratory whether a component passes a limit. The measured value is close to the boundary. The laboratory must distinguish the numerical result from a conformity conclusion and apply the relevant agreed or prescribed decision rule.

Uncertainty informs the technical interpretation and associated decision risk. The laboratory should not invent an acceptance rule after seeing whether the result is commercially convenient. Discuss the reporting and decision needs during contract review, then retain a consistent basis for the conclusion.

Evidence that makes the evaluation understandable

Explore each numbered part

Common nonconformities with clause 7.6

Reviewers meet a familiar set of problems when they look at uncertainty work.

• The reported uncertainty is lifted from a reference certificate and presented as the whole answer.
• An old evaluation is reused after a change of method, equipment or material with no review.
• Significant contributions, such as sampling or environment, are left out because data are hard to collect.
• Calculation spreadsheets are uncontrolled, so formula changes go unnoticed.
• Only senior staff understand the evaluation, and the people using it cannot explain its limits.