ISO/IEC 17025 Measurement Uncertainty — Clause 7.6 Budget Explained
Clause 7.6 of ISO/IEC 17025:2017 makes measurement uncertainty a core competence for accredited laboratories. Every result that can affect decision-making must carry an uncertainty statement. This page walks through what Clause 7.6 requires, how to build a GUM budget, and the mistakes assessors flag.
Clause 7.6 — the requirement
Clause 7.6.1 requires the laboratory to identify the contributions to uncertainty, and to the extent relevant, estimate measurement uncertainty for all calibrations and all types of tests where they affect conformity to a specification. Clause 7.6.2 requires the uncertainty to be evaluated using recognized methods, and Clause 7.6.3 requires it to be reported appropriately.
Building a GUM budget
A defensible budget: identify the input quantities (repeatability, resolution, calibration uncertainty of reference standards, environmental effects, etc.), assign a probability distribution to each, calculate the combined standard uncertainty via the law of propagation of uncertainty, then expand by a coverage factor (usually k=2) for a 95% confidence interval. The GUM is the recognized method assessors expect.
Where budgets fail assessment
Typical deficiencies: using Type A only (repeatability) and ignoring the dominant contributor (the reference standard’s calibration uncertainty); omitting environmental and resolution terms; and stating a single significant figure that over-states precision. Clause 7.6.3 also requires the uncertainty to be reported with enough information for the customer to reproduce it.
Frequently asked questions
Does ISO/IEC 17025 require uncertainty for every test?
Clause 7.6.1 requires it for all calibrations and for tests where uncertainty affects conformity to a specification or is relevant to the validity of the result. Where it is not relevant to state, the laboratory must justify that.
What is the difference between Type A and Type B uncertainty?
Type A is evaluated by statistical analysis of repeated observations (e.g. repeatability). Type B is evaluated by other means — manufacturer specs, calibration certificates, handbooks. A complete budget needs both.
What coverage factor should we use?
For a 95% confidence level a coverage factor of k=2 is typical when the effective degrees of freedom are sufficient. The value should be stated with the expanded uncertainty.