How to read and review a calibration certificate

A calibration certificate records what was measured, against which reference, with what uncertainty and with what result, on one date. Reviewing it means answering four questions: is this the right document for the right instrument, is the traceability and accreditation real, is the uncertainty small enough for your tolerance, and what happened before and after any adjustment? This guide walks through each part of a certificate, using an annotated example, and ends with a printable review checklist.

What a calibration certificate is (and what it isn’t)

Calibration compares an instrument’s indications with reference standards of known value and uncertainty, and documents that relationship (VIM 2.39). It is not the same as adjustment, which changes the instrument so it reads closer to the reference. A calibration service can include an adjustment, but the adjustment is a separate operation, and the instrument must then be calibrated again (VIM 2.39 Note 2; 3.11 Note 3). That is why you can get two sets of results, before and after (ISO/IEC 17025, clause 7.8.4.1).

Not every document that arrives with an instrument is a calibration certificate. The paperwork usually falls into one of these types:

Anatomy of a calibration certificate: an annotated example

ISO/IEC 17025 sets the minimum content of a calibration certificate. Clause 7.8.2.1 lists what every lab report needs (identification, customer, method, item, dates, results, authorization). Clause 7.8.4.1 adds the calibration-specific items: measurement uncertainty, the conditions that influence the results, a traceability statement, results before and after any adjustment or repair, and, when relevant, a statement of conformity (summarized in NIST’s 17025 crosswalk and SOP 1).

The example below is a fictitious certificate for a digital pressure gauge. It follows that structure, with numbered notes. Select a number to jump to its note.

Two details in this example are easy to miss on a real certificate. The as-found reading at 300 psi was outside the ±0.75 psi tolerance, and the lab then adjusted the gauge. The as-left results look perfect, but the as-found row is the one that tells you whether the parts you measured with this gauge since its last calibration were measured correctly.

How to read the results table

Results tables vary from lab to lab, and the column names are a common source of confusion. Map each column to one of these:

• Nominal (test point, set point): the target value the lab aimed for.
• Reference value: what the reference standard measured. This is the “true” value for the comparison.
• Instrument reading (unit under test, UUT): what your instrument indicated.
• Error (deviation): reading minus reference.
• Tolerance and result: the limits used for pass/fail, and the verdict.

If a certificate shows only nominal values and readings, ask the lab for the reference values. Without them you can’t recompute the error.

The sign matters. Error = indication − reference value, and a correction is the same size with the opposite sign: it is what you add to a reading to compensate for a known systematic error (VIM 2.16, 2.53; GUM 3.2.3 and B.2.23 Note 1). If a 15 mm gauge block measures 14.998 mm on your micrometer, the error is −0.002 mm and the correction is +0.002 mm.

As found and as left. ISO/IEC 17025 requires results before and after adjustment or repair “if available” (clause 7.8.4.1). If the lab did not adjust anything, one set of results is normal. If it adjusted the instrument and only reports as-left data, you have lost the information you need for an out-of-tolerance review. Ask for as-found data when you order the calibration.

Measurement uncertainty: what k=2 means and whether it’s good enough

Every calibration result has an uncertainty. On a certificate it appears as the expanded uncertainty U, which the lab gets by multiplying the standard uncertainty by a coverage factor k. Accreditation policy asks for a coverage probability of approximately 95 % and for the certificate to state k and the coverage probability (Global ACI-TECH-1-009, formerly ILAC P14, §4.3 and §5.2). Under the usual conditions, that means k = 2 (EA-4/02, §5).

Three checks take a minute:

1. U and k are both stated. An uncertainty without its coverage factor and coverage probability is incomplete.
2. U has at most two significant digits, and the result is rounded to the same decimal place (TECH-1-009, §5.3; ANAB AR 2251, §4.7–4.8). “0.123456 psi” is a red flag.
3. U is not smaller than the lab’s CMC. The calibration and measurement capability (CMC) in the lab’s scope is the best uncertainty it can achieve. The uncertainty for your instrument is normally larger, because it includes your instrument’s own contributions (TECH-1-009, §5.4–5.5).

Then the question that matters: is U small enough for your tolerance? A common way to express this is the test uncertainty ratio (TUR): the tolerance divided by the 95 % expanded uncertainty (Global ACI-TECH-1-002, formerly ILAC G8, §2.13); for a two-sided tolerance, the tolerance span divided by 2U (A2LA G136). In the example, the tolerance is ±0.75 psi and U is 0.12 psi, so TUR = 1.50 / 0.24 ≈ 6.2:1.

With Lab B, a reading that “passes” by a comfortable margin could still be outside the tolerance. Any time the uncertainty band crosses a tolerance limit, the pass/fail call carries real risk.

“Pass” isn’t always pass: decision rules and guard bands

When a certificate says pass, in tolerance or meets specification, that is a statement of conformity. ISO/IEC 17025 requires the lab to say which results it applies to, which specification was used, and which decision rule was applied: the rule that sets how measurement uncertainty is taken into account (clauses 7.8.6 and 3.7, as described in TECH-1-002 and A2LA G136). The decision rule should be agreed with you, ideally before the work is done (clause 7.1.3).

The rule changes the verdict for results near a limit:

• Simple acceptance (pass if the result is within tolerance): if a result sits right at the limit, the chance that the instrument is actually out of tolerance can be as high as 50 % (TECH-1-002, §5.1).
• Guarded acceptance: the lab only passes results inside an acceptance limit set inside the tolerance. With a guard band equal to U, the specific risk of a false accept stays below about 2.5 % for a single limit with normally distributed results (TECH-1-002, §6.2, Table 1). The price is more false rejects (TECH-1-002, §6.4).
• Non-binary rules add conditional pass and conditional fail for results inside the guard band (TECH-1-002, §5.2.3). Some labs call this zone indeterminate.

The same result can pass under one rule and fail under another. Whoever receives the statement carries the risk of the rule chosen (TECH-1-002, §1.2). Two things to check:

• The certificate names the decision rule. “Simple acceptance; uncertainty not taken into account” is not a valid decision rule according to A2LA, because a decision rule by definition takes uncertainty into account (A2LA G136).
• The specification used is the one you care about. A lab usually checks the manufacturer’s specification. If your process tolerance is tighter, the lab’s “pass” does not answer your question.

Accredited, traceable, or neither?

Accreditation is how you get confidence in a lab without auditing it yourself. In the US, calibration labs are accredited to ISO/IEC 17025 by six bodies that are signatories for calibration: A2LA, ANAB, IAS, NAC, NVLAP (run by NIST) and PJLA. All of them are signatories of the Global ACI Mutual Recognition Arrangement (Global ACI MRA), which took over from the ILAC MRA on January 1, 2026; you can check them in Global ACI’s signatory list. You will still see the ILAC MRA mark on many certificates during the transition (Global ACI-MRA-006, Annex 1). Three checks:

1. The certificate carries the accreditation symbol (or a clear reference to accredited status) and the accreditation number. Only certificates that do so get the full benefit of the MRA for traceability (ILAC P10, Note 3; ANAB AR 2251, §2.1.2). Accredited labs also issue non-accredited certificates, and results outside the accreditation must be clearly identified (Global ACI-MRA-006, formerly ILAC P8, §9.8).
2. The calibration is in the lab’s scope. Look the lab up in the accreditation body’s online directory and open its scope. The parameter and range of your calibration must be listed, and the uncertainty on your certificate cannot be better than the CMC listed there (A2LA R105; TECH-1-009, §5.5).
3. The traceability statement points to the SI, for example “traceable to the SI through NIST” or another national metrology institute (ANAB AR 2251, §2.3.4). NIST states that it does not certify the traceability of results it did not produce, and that a NIST test report number is not proof of traceability (NIST policy and FAQ).

If you use a non-accredited provider (an instrument manufacturer, for example), you will need other evidence of competence, uncertainty and traceability. ILAC P10 lists what that evidence should cover.

When the as-found data is out of tolerance

An out-of-tolerance (OOT) as-found result means the instrument may have been giving wrong readings since its last good calibration. The certificate does not end the problem. It starts your investigation:

• Find where the instrument was used. This is where good records pay off: which products, lots, tests or other calibrations depended on it since the last in-tolerance result.
• Assess the impact. Compare the size of the error with the tolerances of what was measured. A 0.15 psi error matters for a ±0.2 psi test and not for a ±5 psi one.
• Act on the equipment and on affected product, and record what you decided and why.

Quality standards expect this. ISO 9001 (clause 7.1.5) expects the same kind of review when measuring equipment that provides traceability is found unfit for purpose. For medical devices, ISO 13485 (clause 7.6) explicitly requires the assessment and a record of it. FDA inspects medical device calibration under that clause since the Quality Management System Regulation took effect on February 2, 2026 (FDA Compliance Program 7382.850; MDSAP Audit Approach, Tasks 13–14). It replaced the former 21 CFR 820.72.

For a full impact-assessment method, see our guide to out-of-tolerance calibration results.

Red flags and common myths

Red flags on the certificate itself (several are among the findings A2LA assessors report most often in accredited certificates):

• No measurement uncertainty, or U without k and the coverage probability.
• U with more than two significant digits, or smaller than the lab’s CMC.
• No results at all (“calibrated, meets specifications”).
• Only as-left data after an adjustment.
• No issue date, or no unique certificate number on every page.
• A traceability statement that names only NIST or quotes a NIST test number.
• An accreditation symbol on a calibration that is not in the lab’s scope.
• Test points that don’t cover the range you use (one point for a thermometer you use from 2 to 120 °C).
• Model or serial number that doesn’t match your instrument.

Myths you’ll see repeated online:

Calibration certificate review checklist

Use this sequence each time a certificate arrives, before the instrument goes back into use. ILAC recommends checking calibration results against predetermined limits before you use the equipment again (ILAC G24, §4.10). NIST publishes a lab-side review checklist based on ISO/IEC 17025 clause 7.8 (SOP 1, Appendix B).

1. 1

   Right document, right instrument

   Calibration certificate (not a conformity statement), unique number on every page, your model, serial number and asset ID.

   Mismatch: return to the lab

2. 2

   Accreditation and scope

   Accreditation symbol and number present; calibration and range inside the lab's published scope.

3. 3

   Traceability

   Reference standards identified; traceability statement to the SI.

4. 4

   Results versus your tolerance

   Test points cover your working range; errors compared with your process tolerance, not only the manufacturer's.

5. 5

   Uncertainty and ratio

   U and k stated, two significant digits at most, not below the CMC; TUR acceptable for your use.

6. 6

   Statement of conformity

   Specification and decision rule named; results near the limits flagged.

7. 7

   As-found results

   Any out-of-tolerance as-found result opens an impact assessment.

   Out of tolerance: start the OOT process

8. 8

   Accept and record

   Record the acceptance (who, when), update the equipment record and label, and review the interval.

What auditors expect, by standard

Requirements differ by sector. This is what each framework looks for when it comes to calibration certificates from external providers. For paid standards we describe requirements in our own words, based on the official guidance cited.

Does a calibration certificate expire?

No. A certificate reports what the instrument did on the calibration date. When to calibrate again is your decision, based on how the instrument drifts, how much risk an undetected error carries and how heavily it is used (ILAC G24). Labs accredited to ISO/IEC 17025 may not put a recommended interval on the certificate or label unless you agreed to it or a regulation requires it (clause 7.8.4.3). If your certificates show a due date, it is usually because your purchase order asked for one.

See our guide to setting calibration intervals and the TUR calculator.