Calibration Procedure Template: What to Include

A calibration procedure is the document that makes a calibration repeatable — the thing that means two technicians calibrating the same micrometer six months apart do it the same way and get comparable numbers. It is also one of seven conditions UKAS sets for in-house calibration, and the item most often either missing or reduced to a page of bullet points.

This guide sets out a section-by-section structure you can work from, and — more usefully — what each section has to say to satisfy the uncertainty, decision-rule and reporting requirements that UKAS guidance places on the output.

The requirement itself is short. UKAS policy TPS 41 section 4.5(d) lists among its conditions for in-house calibration: "a controlled and documented procedure for each type of calibration." Three words carry the load — controlled, documented, and each type.

One Procedure Per Type, Not Per Instrument

"Each type of calibration" is the scoping rule, and it is more generous than people assume. One procedure covers every 0–25 mm external micrometer in the building, because the method, the reference standards and the uncertainty budget are identical across them.

You need a new procedure when any of those three change:

• The method changes — comparing against gauge blocks is not the same procedure as comparing against a calibrated setting ring.
• The reference equipment changes — a different reference brings a different imported uncertainty.
• The uncertainty evaluation changes — different contributions or divisors mean a different budget.

So external micrometers and dial test indicators need separate procedures even though both are dimensional. But you do not need forty procedures for forty micrometers.

If you are still deciding whether to perform these calibrations at all, in-house vs outsourced calibration covers the other six conditions that come with the decision.

The Structure

Eleven sections. Adapt the numbering to your own document conventions; the content is what matters.

Three of these are where procedures usually thin out to nothing useful: 7, 8 and 10.

Section 6: how many readings

Worth calling out because it is frequently vague. If the procedure says "take readings" without a number, the Type A repeatability term in your uncertainty budget has no defined basis — the divisor is √n, and n has to be a known quantity. Fix the number in the procedure and the budget stays valid.

Section 7: The Uncertainty Section

This is not a place to write "uncertainty is considered." The procedure should point at the uncertainty budget for this calibration type and state the reported uncertainty that results.

M3003 provides a step-by-step procedure for the evaluation in its section 7, which cross-references a worked example in Appendix K — a reasonable structure to follow when writing this section for the first time. The mechanics of building the budget, including the divisor table and a fully worked example, are in how to build an uncertainty budget, and UKAS M3003 is the document behind it.

Two things this section should nail down:

• Which contributions are included, so a technician does not have to re-derive the budget each time.
• The reported uncertainty for this calibration type, so what appears on the output is consistent.

The budget depends on the reference standard named in section 2, which is why changing the reference means revisiting the procedure — see calibration standards and reference equipment for how the reference's own uncertainty sets the floor.

Section 8: The Decision Rule

If your calibration produces a pass or a fail — and most in-house calibration does — the procedure has to say how the pass/fail is decided, not just what the tolerance is. UKAS LAB 5 section 3.5 is explicit about the output requirement:

"Unless the decision rule is implicit in a specification quoted, the decision rule used to make statements of conformity shall be described in the certificate or report. This is required by clause 7.8.6.2 (c) in ISO/IEC 17025:2017."

Since the certificate or record is generated from the procedure, the procedure is where the rule belongs. Leave it out and each technician effectively invents one — usually simple acceptance, unknowingly, with a 50% chance of a wrong call at the tolerance limit.

LAB 5 section 3.4 adds that where conformity statements are made, "the measurement uncertainty shall be taken into account", pointing to LAB 48, M3003 and ILAC G8. It also covers the case where conformity is stated but results and uncertainties are omitted: "then there should be a statement that describes the calibration points for which conformity is ascribed." A bare "PASS" with no indication of which points were tested is not enough.

State the rule concretely. Not "results are compared to tolerance" but something with numbers in it — a guard band width, or simple acceptance with its capability-index constraint. Decision rules and guard banding covers both forms and how to calculate them.

Section 9: What the Output Must Carry

The procedure determines what ends up on the record, so the reporting requirements are effectively procedure requirements.

LAB 5 section 3.2: it is "necessary for the interpretation of calibration results for calibration certificates to include the measurement uncertainty and/or a statement of conformance with an identified metrological specification or clauses thereof."

LAB 5 section 3.3 gives the uncertainty statement wording for the normal case:

"The reported expanded uncertainty is based on a standard uncertainty multiplied by a coverage factor k = 2, which for a normal distribution corresponds to a coverage probability of approximately 95%. The uncertainty evaluation has been carried out in accordance with UKAS requirements."

It notes that a coverage factor other than 2 is sometimes required and the statement modified accordingly, and that there are circumstances where the effective degrees of freedom should be reported, with examples in M3003.

There is also a rule about due dates that catches people out. LAB 5 section 3.7:

"Any recommendation on the calibration interval must have been agreed with the customer, therefore it shall be clear from the certificate that if a calibration due date is reported it is at the customer's request, e.g. 'Customer requested calibration due date'."

The interval belongs to whoever owns the equipment, not to whoever performed the calibration. If your in-house procedure stamps a due date, the interval should trace back to your own documented interval decision — see determining calibration intervals — rather than being invented at the bench.

LAB 5 also has presentational recommendations in section 3.1, including that dates appear in day/month/year format. Small, but it removes an ambiguity worth removing.

Section 10: When the Result Fails

The commonest gap. A procedure that describes only the happy path leaves the technician deciding what to do about a failure, on the day, under time pressure.

This section should cover: quarantining the instrument, who is notified, how the instrument's status is marked, and — the part that gets missed — the assessment of whether previous measurements taken with it are still valid. The out-of-tolerance procedure covers that assessment in full; the calibration procedure needs to point at it and make clear it is not optional.

Mistakes That Show Up at Audit

Patterns worth checking your own procedures against:

• No stated number of readings — leaves the Type A term in the budget undefined.
• No decision rule — the single most common gap, and LAB 5 §3.5 makes it a reporting requirement.
• Reference standards named by type, not identity — "gauge blocks" instead of the specific set, so nobody can tell which reference a past calibration used.
• Only outcomes recorded, not raw readings — you cannot recompute, review or defend a result you did not keep the numbers for.
• Environmental conditions stated without a tolerance or a soak time — "at room temperature" is not a condition.
• No handling of failures.
• Uncontrolled — no version, or a version on the bench that is not the current one. TPS 41 §4.5(d) asks for controlled, and this is the easiest of the eleven sections to get wrong.
• A procedure that describes a capability you do not have — naming a reference whose own calibration has lapsed, or an environment you cannot actually hold.

The Short Version

• The requirement: "a controlled and documented procedure for each type of calibration" (TPS 41 §4.5(d)).
• Per type, not per instrument. New procedure when the method, the reference equipment, or the uncertainty evaluation changes.
• Eleven sections: scope; reference standards and equipment; environment; personnel; preparation; method; uncertainty; acceptance criteria and decision rule; recording and reporting; non-conforming results; document control.
• Fix the number of readings — the Type A divisor is √n.
• State the decision rule — LAB 5 §3.5 requires it on the output where conformity is stated; the procedure is where it should be set.
• Use LAB 5 §3.3's uncertainty wording, adjusted where k ≠ 2.
• A due date goes on a certificate only at the customer's request, labelled as such (§3.7).
• Include the failure path, and point it at the out-of-tolerance assessment.

For related practical assets, the free downloads library has a calibration schedule template and an out-of-tolerance decision flowchart.

How CalProof Fits

A procedure and the calibrations performed under it usually live apart — the procedure in a document system, the results in a spreadsheet or a folder of certificates. That separation is what makes an audit trail hard to reconstruct: you can show the procedure, and you can show a result, but showing which version of the procedure produced a given result takes archaeology.

CalProof holds each calibration event against its instrument, with an in-tolerance/out-of-tolerance result, the certificate, and an optional uncertainty field where you record the uncertainty stated on the certificate (for example the expanded uncertainty and coverage factor) — it stores what you enter, it does not compute or propagate an uncertainty budget. Reference standards are tracked as instruments in the same register, with their own intervals and due dates, and out-of-tolerance events get a recorded workflow rather than an email thread.

GBP pricing from £29 a month. No per-user fees on Pro and above. UK data hosting. No long-term contract.

To see the shape of the output, the sample audit pack is ungated and downloads as a single PDF.

Sources

• UKAS LAB 5 Edition 5, January 2025 — Reporting calibration results
• UKAS TPS 41 Edition 6, December 2022 — UKAS policy on metrological traceability
• UKAS M3003 Edition 6, March 2024 — The expression of uncertainty and confidence in measurement
• UKAS LAB 48 Edition 5, July 2024 — Decision rules and statements of conformity

This guide draws on UKAS LAB 5 Edition 5 (January 2025), TPS 41 Edition 6 (December 2022) and M3003 Edition 6 (March 2024) as published; section references are to those editions. The clause attribution in LAB 5 §3.5 is LAB 5's own characterisation of ISO/IEC 17025:2017. The structure suggested here is a starting point, not a compliance guarantee — the content your procedures need depends on your calibrations, your quality system and your assessor's expectations. Verify against the current published editions and your UKAS assessor or certification body. This is not legal or compliance advice.