PHARMA LAB · PL-05-022

How to Review a Calibration Certificate: A QA Checklist

A complete certificate may still fail to cover intended use. Review identity, results, uncertainty, decision rules and revisions with a practical QA checklist.
Technical illustration of an unlabelled calibration results table, a magnifying glass and a laboratory thermometer with a probe.

QA review of a calibration certificate must answer two different questions: does the document reliably describe the service performed? Are its results relevant to the measurements for which the instrument will be used? A formally complete document can pass the first check and fail the second.

The checklist below accompanies an invented results table. It is not a facsimile accredited certificate and does not replace the requirements of company procedures, contracts or applicable standards.

1. Connect issuer, document and instrument

Check the issuing laboratory, unique document identifier, pages and attachments, authorization for issue and revision. The calibration certificate is issued by the laboratory performing and reporting the service; the accreditation body accredits that laboratory and does not become the author of each report. Distinguish certificate number from accreditation number.

Compare serial number, internal identifier, model, probe, channel, cables and configuration with the inventory and order. Read activity, issue and intervention dates separately. A printed “next calibration” date does not replace the user's review of the calibration interval.

For a service requested under accreditation, check the relevant scope and declared limitations, including location and excluded activities. A symbol on the first page does not remove notes attached to individual results. Provider qualification and acceptance of this document remain distinct checks.

2. Compare method, points and conditions with the request

Identify what was measured: indication of the complete chain, probe response or only electrical input simulation. Check method, points, operating modes, relevant environmental conditions and references used or the traceability statement. Seek clarification when these elements do not allow interpretation of the result.

The instrument's nominal range is not the range actually examined. Points far from intended use do not authorize automatic extrapolation; interpolation also needs a technical basis and uncertainty assessment. Numerous points do not compensate for an incorrect configuration or a different channel.

3. Read units, signs and uncertainty together

Identify the reported quantity: assigned value, indication, error and correction are not interchangeable. Below, error is defined as e = indication − reference; the additive correction is C = −e. Other formats require reading the laboratory's definition before applying a sign.

Expanded uncertainty U describes the result together with coverage factor k and the stated coverage probability. It is not a tolerance, correction or guaranteed maximum error. When the model uses U = k·u, U = 0.04 °C with k = 2 corresponds to u = 0.02 °C. Do not divide again if a value is already expressed as standard uncertainty.

Check consistency between absolute and relative units, decimal places, rounding and the assignment of uncertainty to the correct result. k = 2 alone does not establish an exact universal coverage probability. Calibration uncertainty does not automatically include subsequent drift, transport, environment and use; these contributions belong in the assessment of the measurement in use.

4. Separate chronology from conformity statements

Read as-found and as-left conditions alongside the intervention sequence. A result after adjustment does not demonstrate the initial state; a missing initial result does not mean zero error. If measurements were not taken before repair, request a faithful description of the available evidence, without invented retrospective reconstruction.

For a conformity statement, identify the specification, limits, covered results and decision rule. “PASS” against a manufacturer's specification does not authorize every process. Different rules may give different outcomes near a limit; the choice must be defined before work, consistent with applicable requirements, and not changed afterwards to obtain a favourable outcome.

5. Original QA review checklist

FieldQA questionPossible anomalyAction
Identity and revisionIs this the correct instrument and complete document?Conflicting serial number or missing attachmentHold final review and request issuer clarification
Service and scopeDo activity and location match the request?Unplanned service outside scopeAssess the discrepancy with the provider qualification owner
Method and pointsDo they support the measurements of interest?Points outside the operating rangeRequest relevant evidence or measurements
Results and UAre signs, units, k and coverage clear?Correction confused with errorCheck definitions and calculation without editing the report
InterventionsWhich state does each table describe?Missing as-found data after repairRecord the limitation and assess its impact
ConformityWhich criterion and rule produce the outcome?PASS with no intelligible scopeObtain the basis of the statement
ClosureWho accepts the document and decides on use?Administrative signature treated as technical releaseRecord review and authorization for use separately

Adapt the checklist by adding the order reference, controlled evidence, reviewer, date and open actions. A relevant uncertainty in interpretation remains visible until resolved; it does not disappear because every other box is complete.

6. Simulated case: correct results, uncovered operating range

A thermometer is intended for measurements between 2 and 8 °C. The document reports only these as-left results; all temperatures and U are in °C, k = 2:

ReferenceIndicationError eCorrection CU
20.0020.08+0.08−0.080.04
40.0040.10+0.10−0.100.04
60.0060.12+0.12−0.120.04

The service states conformity only at measured points to an illustrative ±0.20 °C limit, accepting when |e| + U ≤ 0.20 °C. The comparisons are 0.12, 0.14 and 0.16 °C: all favourable. At the first point, 20.08 + (−0.08) = 20.00 °C shows application of the defined correction, not elimination of uncertainty.

These results do not establish performance between 2 and 8 °C. QA compares order and certificate, requests relevant measurements and a justification of coverage, and withholds authorization for that use until sufficient evidence is available. It does not apply −0.08 °C throughout the operating range or assume a favourable initial state.

If the document needs correction, request revision from the issuer: retain the original, new version, reason and link between documents. Do not rewrite the provider's PDF. See as-found/as-left results and provider qualification; return to the Calibration and Metrology hub.

7. Sources and application limits

Checked on 1 October 2026. The checklist, table and numerical limit are original simulations, not universal GMP requirements.

  1. ISO/IEC 17025:2017: catalogue, confirmed in 2023; full text not consulted. Check reporting requirements in a controlled copy.
  2. ILAC-P14:09/2020: complete document available; section 5 consulted on uncertainty reporting and coverage. Accreditation policy.
  3. ILAC-G8:09/2019: sections 2, 4 and 7 consulted on statements and decision rules; no single rule for all situations.
  4. Cofrac LAB REF 02, revision 15: applicable from 1 March 2026; sections 7.8.2, 7.8.4, 7.8.6 and 7.8.8 consulted. Requirements of that accreditation system.
  5. Accredia, calibration laboratories: full institutional page consulted, no visible revision date; issuer and certificate limitations.
Technical content for informed decisions; it does not replace the approved procedure, applicable requirements or the instrument manual.

Continue exploring