Guide
What should a calibration certificate show to prove a supplier's gauges are traceable?
A calibration certificate is the document a calibration laboratory issues for one specific instrument, recording what was measured, the results and their measurement uncertainty, the conditions of the calibration, and the chain of reference standards that makes those results metrologically traceable to the SI, usually through a national metrology institute such as NIST.
By Saola Manufacturing engineering team · Published · Updated
Every supplier audit reaches the same moment: the auditor points at a micrometer on the bench and asks to see its calibration. What comes back is usually a binder or a shared folder of certificates, and the question is whether those documents actually prove that the numbers on your inspection report are right. This guide is about reading that evidence. It covers what ISO/IEC 17025 requires a certificate to contain, how to tell an accredited calibration from a bare “traceable to NIST” stamp, how to judge whether the uncertainty is small enough for your tolerance, what to look for in the supplier's gauge register, and what to write into the RFQ and purchase order. Gauge classes and how a go/no-go thread gage is used are covered in the thread gage guide; the wider supplier audit is in the supplier vetting checklist.
On this page
What must a calibration certificate contain under ISO/IEC 17025?
Per ISO/IEC 17025:2017 clause 7.8, a calibration certificate must identify the laboratory, the customer, the item and the method, and must report the results with their measurement uncertainty, the conditions that influence them and a statement of how they are metrologically traceable. A document missing any of those is a service receipt, not evidence an auditor can rely on.
The general reporting requirements in clause 7.8.2 apply to every test and calibration report; clause 7.8.4 adds the items specific to calibration. The table below lists the ones that matter when you are judging a supplier's gauge, and the red flag that tells you the item is missing or meaningless. Read the certificate against the actual gauge, not against the folder index: the serial number or asset ID on the certificate has to match the marking on the instrument you are holding.
| Item on the certificate | What it should say | Red flag |
|---|---|---|
| Unique certificate ID, page numbering, clear end | An ID on every page and “page x of y” or an end-of-report mark | Loose pages that could belong to any certificate |
| Laboratory name, address and place of calibration | The lab, and whether work was done at the lab or on site | No address, or a trading name you cannot find on any accreditation scope |
| Item identification and condition | Make, model, serial number or asset ID, range and resolution; condition on receipt | “Digital caliper” with no serial or ID that ties it to one instrument |
| Method | The procedure used, often referencing a product standard such as ISO 13385-1 for calipers or ISO 3611 for micrometers | No method named |
| Date of calibration and date of issue | Both dates; a due date only if agreed with the customer | Every certificate in the folder issued the week before the audit |
| Results with units | Measured values at stated test points across the range | “PASS” or “within tolerance” with no data on a gauge used near the part limit |
| As-found and as-left results | Readings before and after any adjustment or repair | Only as-left data, which hides how far the gauge had drifted while in use |
| Measurement uncertainty | Expanded uncertainty with its coverage factor and confidence level | No uncertainty at all, or an uncertainty smaller than the lab's accredited capability |
| Environmental conditions | Temperature, and any other condition that affects the result | Blank for a dimensional calibration |
| Traceability statement | Traceable to the SI through named reference standards and their calibrating body | A NIST test number offered as the only proof |
| Statement of conformity and decision rule | If the lab says “in tolerance”, the specification and the decision rule applied | “In tolerance” with no tolerance named |
| Authorization | Identification of the person who authorized the certificate | No authorizing person identified, or a sign-off by the supplier's own staff in place of the lab's |
Two items deserve a second look. The standard does not allow the laboratory to put a recommended recalibration interval on the certificate unless the customer agreed to it, so the interval is the supplier's decision and belongs in the supplier's register. And the conformity statement is optional: a certificate can be perfectly valid with data and uncertainty and no pass/fail verdict, in which case the supplier is expected to compare the data with its own acceptance limits and record that it did.
How do you tell an accredited calibration from one that only claims traceability?
An accredited calibration carries the accreditation body's symbol and the lab's accreditation number, and the instrument type and range appear on the lab's published scope of accreditation. A certificate without accreditation can still be valid, but the whole burden of proving the chain then falls on the supplier and on you.
Checking accreditation takes a few minutes and is the single most useful thing a buyer can do with a calibration folder. First, identify the accreditation body named on the certificate and confirm it appears on the ILAC MRA signatory list for calibration. In the US that is usually A2LA, ANAB or NVLAP; in the UK, UKAS; in Canada, SCC; in Australia, NATA; and in the supplier's own country, its national accreditation body. Accreditations granted by any MRA signatory are recognized by the other signatories as equivalent, so a certificate from a lab in Vietnam or Mexico is not weaker for being foreign, provided the accreditation is real.
Second, download the lab's scope of accreditation from the accreditation body's own website, not from the lab. The scope lists each parameter, the range and the smallest uncertainty the lab may claim, called the calibration and measurement capability (CMC). Confirm that the instrument type is on it and that the range covers the gauge: a scope that covers micrometers and gauge blocks may say nothing about height gauges or thread rings.
Third, compare the uncertainty on the certificate with the CMC. Under the ILAC policy on uncertainty in calibration (ILAC P14), a reported uncertainty cannot be smaller than the CMC on the scope. Finally, look for the accreditation symbol on the certificate itself. Accredited labs often do work outside their scope as well, and under ILAC P8 those results must not be presented as accredited; a certificate with neither the symbol nor a stated reference to the accreditation is not an accredited calibration, even if the lab is accredited for other things.
What does “NIST traceable” actually prove on a supplier's calibration records?
On its own, very little. NIST states that it does not certify the traceability of anyone else's measurements, that supporting a traceability claim is the job of whoever provides the result, and that judging the claim is the job of the user.
The NIST policy on metrological traceability adopts the definition from the International Vocabulary of Metrology: a result that can be related to a reference through a documented unbroken chain of calibrations, each contributing to the measurement uncertainty. The last half of that sentence is the part most supplier records miss. If any link in the chain has no stated uncertainty, the chain is broken, however many times the word “traceable” appears on the page. The policy also says plainly that traceability alone does not mean fitness for purpose, because the uncertainty may still be too large for the measurement you need.
The NIST traceability FAQ addresses the most common shortcut directly: a NIST test report number is an administrative reference and should not be used or required as the sole proof of traceability. A certificate that lists a NIST test number and nothing else has told you where a reference standard was once calibrated, not that this gauge is fit to inspect your part.
For an overseas supplier there is a second point. The chain does not have to pass through NIST. A lab in Vietnam will normally trace to its own national metrology institute or to another one, and national institutes that take part in the CIPM mutual recognition arrangement, coordinated by the BIPM, which also publishes the VIM and the GUM, recognize each other's calibration capabilities. Write “traceable to the SI” rather than “traceable to NIST” in your requirements unless a contract or regulation names NIST, and then ask how the chain gets there.
Is the calibration uncertainty small enough for your part tolerance?
Compare the uncertainty on the certificate with the tolerance of the features the gauge is used to check. ANSI/NCSL Z540.3 accepts a test uncertainty ratio of 4:1 or better when the gauge is calibrated, and the shop-floor rule of ten asks for a gauge resolution of about one-tenth of the part tolerance when it is used.
The 4:1 figure comes from ANSI/NCSL Z540.3-2006, which requires that the probability of a false accept in calibration not exceed 2% per ANSI/NCSL Z540.3, and accepts a test uncertainty ratio of 4:1 or better where estimating that probability is not practicable. That ratio protects the calibration of the gauge. It does not by itself make the gauge suitable for a feature on your part, which is the question you actually care about.
A worked example shows why. A bore with a total tolerance band of 0.020 mm, under the typical rule of ten, needs an instrument that resolves 0.002 mm or better. A typical digital caliper reads in 0.01 mm steps, so even with a flawless certificate it is the wrong instrument for that bore; a micrometer, bore gauge or air gauge is the right one, and its certificate is the one you need to see. When you review records, start from the critical features on the drawing and work back to the gauges, rather than starting from the folder.
Look next at the as-found data. If a gauge arrived at the lab outside its own tolerance, every part measured with it since the previous good calibration is in question. A competent supplier records that event, works out which lots were affected and decides whether to recall or re-inspect. Ask to see one example of that reaction record; a supplier that has never found a gauge out of tolerance has usually never looked at as-found data.
Finally, check how any pass/fail statement was reached. Clause 7.8.6 of ISO/IEC 17025 requires the lab to state the decision rule, and the ILAC guidance series (ILAC G8 on decision rules) explains the options. Under simple acceptance a reading just inside the limit passes even though the uncertainty straddles the limit; under guard banding it does not. Either can be right, but you should know which one was used before you accept a gauge that sits near its own limit.
What should you check in the supplier's gauge register, beyond the certificates?
Check that every gauge used on your part has a unique ID that appears on the gauge, in the register and on its certificate, a due date that had not passed when your parts were measured, and a recorded reaction whenever a gauge was found out of tolerance. Certificates prove one calibration; the register proves the system.
The register requirements come from the quality-system side rather than the laboratory side. ISO 10012:2026, which replaced the 2003 edition, sets out a full measurement management system, and a supplier whose ISO 9001 registration is in progress or not yet started can still run a register that meets the measurement traceability clause of ISO 9001 (clause 7.1.5.2 in the 2015 edition) in full: identified equipment, calibration or verification at intervals against traceable standards, protection from adjustment and damage, and an assessment of earlier results when a gauge is found unfit.
The quickest audit is to work from the floor to the paper. Pick five gauges off the benches at random, including at least one thread gage and one height gauge, and trace each one to the register and to its certificate. Gauges with no ID, personal tools in a toolbox, and instruments labeled “reference only” that are plainly being used to accept parts are all findings.
Working gauges do not all need to go to an outside lab. Calibrating calipers and micrometers in-house against gauge blocks, setting rings or master plugs is normal and acceptable, provided there is a written procedure, a controlled environment near the reference temperature of 20 °C per ISO 1, and accredited certificates for the reference standards themselves. In that case the in-house record will not look like an ISO/IEC 17025 certificate, and it does not need to; what you check is the procedure, the data and the certificate of the block set behind it.
Intervals are the supplier's call. A 12-month interval is a typical starting point for reference standards and many hand gauges, shortened for gauges that see heavy use or show drift and lengthened only on evidence, using methods such as those in the ILAC G24 guidance on recalibration intervals. Thread plug gages wear fastest on the GO member, which is why high-volume tapped holes justify a shorter interval or a wear check between calibrations.
Common red flags in a register: every certificate dated within a few days of your audit notice; identical results on several different instruments; a manufacturer's certificate from the day the gauge was bought that has never been renewed; due dates that passed during your production run; and a register kept as a spreadsheet nobody can show a revision history for.
What should the RFQ, purchase order and quality agreement say about gauge calibration?
Say which measuring equipment must be calibrated, against what, by what kind of lab, and what happens when a gauge is found out of tolerance, and ask for the evidence with the first-article report rather than after a dispute. One purchase order clause and a short quality-agreement paragraph cover it.
A purchase order clause along these lines covers it:
- “Measuring equipment used to accept characteristics on this order shall be identified, calibrated at defined intervals and traceable to the SI. Reference standards shall be calibrated by a laboratory accredited to ISO/IEC 17025 by an ILAC MRA signatory, with the item within its scope of accreditation.”
- “Calibration certificates shall report as-found and as-left results and expanded uncertainty. The supplier shall notify the buyer in writing when a gauge used on this order is found out of tolerance, with the lots affected.”
- “The first-article inspection report shall list, for each characteristic, the gauge ID and its calibration due date.”
The last line is the one that pays off. It ties each balloon on the drawing to a specific instrument, so a disputed reading can be traced to a certificate in minutes; the ballooned drawing guide and the first article inspection guide show how that report is laid out.
Do not over-specify. Requiring an accredited outside calibration for every caliper on the floor adds cost without adding much assurance, because a well-run in-house check against accredited gauge blocks gives the same traceability. Reserve the accredited-lab requirement for reference standards, thread gages and special gauges, and for any instrument that checks a critical characteristic. If you also need to know how much of the variation in the readings comes from the gauge itself, that is a separate gauge study; the Cpk vs Ppk guide explains how measurement error feeds into capability numbers.
What inspection evidence does Saola publish today, and what should you ask for?
Saola's quality page lists a first-article inspection report on every new part and a dimensional inspection record with each lot, and states that ISO 9001 certification is in progress. Its published pages do not describe a calibration register or name a calibration laboratory, so ask for gauge calibration records in your RFQ and judge them with the checks above.
The gauging the site does describe is specific. The CNC turning page says thread fit is checked with go/no-go gauges before a lot ships, and that ANPT and NPT threads to ANSI/ASME B1.20.1 use gauges the plant already owns. The factory page's sample inspection plan names calipers, go/no-go pin gauges, dedicated fixture gauges and a surface plate, and its floor photos show a height gauge on a granite plate. Those are exactly the instruments the register checks above apply to, so the request is straightforward: name the characteristics that matter on your drawing, and ask for the gauge ID and calibration evidence behind each one with the first-article report. Send the drawing and that request together through the quote page.
Frequently asked
Does every gauge need an ISO/IEC 17025 accredited calibration?
No. Quality-system standards require measuring equipment to be calibrated or verified against traceable standards, not that every instrument visit an accredited lab. Accredited calibration matters most for reference standards, thread gages, special gauges and instruments that accept critical characteristics; working hand gauges can be checked in-house against those references under a written procedure.
Is the certificate that came with a new gauge good enough?
It can be, if it was issued by an accredited lab, identifies the individual instrument and reports data with uncertainty. A generic inspection slip that says the gauge passed final inspection at the maker is not a calibration certificate. Either way, the gauge then needs to enter the supplier's register with an interval of its own.
Who decides the calibration interval, the lab or the supplier?
The supplier. ISO/IEC 17025 keeps recommended intervals off the certificate unless the customer asked for one, because the right interval depends on how hard the gauge is used and how much it drifts. The supplier should be able to explain how each interval was set and when it was last reviewed.
What is the difference between calibration and a gauge R&R study?
Calibration compares a gauge with a traceable reference and finds its bias and uncertainty. A gauge repeatability and reproducibility study measures how much readings scatter when the same parts are measured repeatedly by different people. A gauge can pass one and fail the other, so a supplier's capability data needs both.
Should a calibration certificate be in English?
Ask for English, or bilingual, in the quality agreement. A certificate in another language is not invalid, but an auditor who cannot read the traceability statement and the uncertainty cannot verify them, and a translation prepared by the supplier is not a substitute for the lab's own document.
Sources
- ISO/IEC 17025, General requirements for the competence of testing and calibration laboratories (clauses 7.8.2, 7.8.4, 7.8.6), ISO and IEC, 2017
- NIST Policy on Metrological Traceability, NIST, accessed October 2026
- Metrological Traceability: Frequently Asked Questions and NIST Policy, NIST, accessed October 2026
- ILAC MRA and signatories, International Laboratory Accreditation Cooperation, accessed October 2026
- ILAC policy series, including ILAC P14 on uncertainty in calibration, ILAC, accessed October 2026
- ILAC guidance series, including ILAC G8 on decision rules and ILAC G24 on recalibration intervals, ILAC, accessed October 2026
- JCGM publications: International Vocabulary of Metrology (VIM) and Guide to the Expression of Uncertainty in Measurement (GUM), BIPM, accessed October 2026
- National Voluntary Laboratory Accreditation Program (NVLAP), NIST, accessed October 2026
- ISO 10012:2026, Quality management: Requirements for measurement management systems, ISO, 2026 (replaces ISO 10012:2003)
- ANSI/NCSL Z540.3-2006, Requirements for the Calibration of Measuring and Test Equipment, NCSL International, 2006