Summary
In short
- As-found readings are the whole point. Adjusting first and recording only as-left destroys the evidence needed to assess what the instrument was doing before.
- An out-of-tolerance finding invalidates every measurement since the last confirmed good calibration and requires an impact assessment across that period.
- The tolerance that matters is the one your decision requires, not the manufacturer specification. An instrument well within spec may still consume too much of the tolerance you are measuring against.
- Intervals should come from stability history and consequence. An out-of-tolerance rate of zero across a fleet usually means intervals are too short, which is expensive rather than safe.
- Measurement uncertainty belongs on the certificate. A reading reported without it cannot support a pass or fail decision near a limit.
- Instruments must be linked to what they measured. Without that link, the impact assessment becomes a manual reconstruction under time pressure, which is where scope errors happen.
What it is
What it is
What is a calibration record?
A calibration record documents the comparison of a measuring instrument against a traceable reference standard: what standard was used, what the instrument read before any adjustment, what it read after, the measurement uncertainty, whether it met the required tolerance, and when it is next due.
What does traceable mean?
That the reference standard used is itself linked, through an unbroken chain of comparisons each with stated uncertainty, back to a national or international standard. A certificate that does not identify the reference and its traceability is not evidence of traceability, and calibrating against an uncalibrated reference provides none.
What is the difference between calibration and verification?
Calibration compares the instrument to a standard and reports the deviation. Verification confirms the instrument meets a specified requirement. An instrument can be calibrated with a large deviation reported, which is useful information, and fail verification against your tolerance. ISO 9001 permits either, provided the basis is defined.
When to use it
When to use it, and when not to
The calibration record covers one calibration event on one instrument. Several adjacent activities are frequently pushed into it and belong elsewhere.
Use it for
- Scheduled calibration of measuring and test equipment against traceable standards
- Calibration after repair, adjustment, damage or suspected malfunction
- In-house verification checks between formal calibrations, where the method is defined
- Receipt of a new instrument, establishing its baseline before first use
- Any instrument returning from a period of storage or removal from service
Not for
- The instrument register itself, which holds identity, location, tolerance requirement and interval
- Measurement system analysis and gauge R&R, which assess whether the measurement method is capable for its purpose
- The impact assessment triggered by an out-of-tolerance finding, which is its own record with its own scope decision
- Maintenance and repair records for the instrument, which document work rather than measurement
- Method validation, which establishes whether the test method is fit for purpose
Standards
What it is built against
Calibration requirements are consistent across quality standards and become sharply more prescriptive where product release depends on the measurement.
| Clause | Requirement | Where it lands |
|---|---|---|
| ISO 9001 cl.7.1.5.1 | Monitoring and measuring resources suitable, and maintained to ensure continued fitness | Header |
| ISO 9001 cl.7.1.5.2 | Calibrated or verified against traceable standards, identified, safeguarded, and action taken when found unfit | Standards used, Outcome |
| ISO 9001 cl.7.1.5.2 | Validity of previous results assessed when equipment is found unfit for purpose | Outcome |
| ISO/IEC 17025 cl.6.4 | Equipment calibration programme with traceability and stated measurement uncertainty | As found readings |
| ISO/IEC 17025 cl.6.5 | Metrological traceability established through documented unbroken chain of calibrations | Standards used |
| 21 CFR 211.68 | Written calibration programme with limits for accuracy and precision, and action when limits are exceeded | Outcome |
| 21 CFR 820.72 | Calibration procedures with accuracy and precision limits, remedial action and records identifying the person | Whole record |
| GFSI schemes | Calibration of equipment critical to food safety and legality, with action on out-of-calibration findings | Outcome |
What it does not cover
- The impact assessment, which determines what product and decisions were affected by an out-of-tolerance instrument and needs its own scope reasoning.
- Gauge R&R and measurement system analysis, which establish whether the measurement is capable for the tolerance it checks.
- The instrument register, which holds required tolerance, interval, location and criticality.
- Method validation, which addresses the test method rather than the instrument.
- Repair records, which document work on the instrument rather than its measurement performance.
Filling it in
Filling it in well
Four fields carry most of the value, and two of them are routinely compromised in the interest of a clean record.
The first action is to measure and record what the instrument reads in the condition it arrived. Only then adjust, and record as-left separately. A record showing only as-left states that the instrument is accurate now, which nobody was asking, and cannot answer the question that matters when a batch is in dispute.
Manufacturer specification describes what the instrument can do. The relevant question is what accuracy the decision depends on. A thermometer accurate to plus or minus one degree may be entirely adequate for a warehouse ambient check and inadequate for a critical control point with a two degree window, and the same instrument passes one and should fail the other.
Every calibration has a measurement uncertainty, and a reading close to the tolerance limit may be inside or outside it once uncertainty is accounted for. Where uncertainty is not stated, near-limit results are decided by convention rather than by evidence, which is a problem that surfaces during a dispute rather than during the calibration.
When an instrument is found out of tolerance, the immediate question is what it measured since the last confirmed good calibration. Where calibration and production records live in separate systems, answering that is a manual reconstruction performed under time pressure with product on hold, which produces both over-scoping and missed batches.
Audit findings
Common audit findings
Calibration findings concentrate on the as-found field and on what happens after a failure.
| Finding | Clause | What fixes it |
|---|---|---|
| As-found readings not recorded; instrument adjusted before measurement. | ISO 17025 cl.6.4 | Record as-found first as a mandatory field; adjustment cannot precede it. |
| Out-of-tolerance finding with no impact assessment on previous results. | ISO 9001 cl.7.1.5.2 | Trigger the assessment automatically on an out-of-tolerance outcome and block closure until it is scoped. |
| Instrument in use past its due date, not quarantined. | 21 CFR 211.68 | Prevent use rather than report overdue status; a report does not stop a measurement. |
| Traceability of the reference standard not evidenced on the certificate. | ISO 17025 cl.6.5 | Require the reference standard identity and its own certificate reference on every record. |
| Tolerance recorded as manufacturer specification rather than the requirement of the decision. | ISO 9001 cl.7.1.5.1 | Define required tolerance per instrument in the register, derived from what it is used for. |
| Measurement uncertainty absent, so near-limit results are decided by convention. | ISO 17025 cl.7.6 | State uncertainty and define the decision rule for results near the limit. |
| Intervals uniform across the fleet with no basis in drift history. | ISO 9001 cl.7.1.5.2 | Set from stability history and criticality; review intervals against actual as-found trends. |
| Calibration performed in-house against an uncalibrated reference. | ISO 17025 cl.6.5 | Traceability requires an unbroken chain; an uncalibrated reference breaks it entirely. |
| No link between the instrument and the batches or decisions it informed. | ISO 9001 cl.7.1.5.2 | Record instrument identity on the measurement so the impact scope can be determined rather than reconstructed. |
| Instrument identification missing or illegible, so status cannot be confirmed at the point of use. | ISO 9001 cl.7.1.5.2 | Label with unique ID and calibration status where the user will see it. |
Worked case
Case in point: the thermometer that passed
A dairy plant calibrated its pasteuriser thermometers annually. At the March calibration a technician found one reading 1.4 degrees low, adjusted it, recorded the as-left result as within tolerance, and marked the calibration passed. The record was complete by the site's own form.
Four months later a customer complaint prompted a review. Quality wanted to know whether the instrument had been reading low before adjustment and by how much, because the critical limit had a two degree window and a 1.4 degree low reading meant product may have been processed below the limit while appearing to pass.
The as-found value existed only in the technician's memory. The site could not establish the magnitude of the error, could not scope which production was affected, and held eleven months of product because the alternative was to guess.
Definitions
Definitions and key terms
- As-found
- The instrument's reading before any adjustment, and the only evidence of its performance during the period it was in service.
- As-left
- The reading after adjustment, describing the instrument's condition going forward.
- Traceability
- An unbroken chain of comparisons to a national or international standard, each with stated uncertainty.
- Measurement uncertainty
- The quantified doubt about a measurement result, required for any decision made near a tolerance limit.
- Out of tolerance
- An as-found result outside the required limits, triggering assessment of every measurement since the last confirmed good calibration.
- Decision rule
- The documented basis for declaring conformity when a result falls near a limit, accounting for uncertainty.
- Calibration interval
- The period between calibrations, set from stability history and consequence rather than by default.
- Reference standard
- The instrument or artefact used as the comparison basis, which must itself be calibrated and traceable.
FAQ
Frequently asked questions
Why are as-found readings so important?+
Because they are the only evidence of what the instrument was doing while it was in use. As-left tells you it is accurate now. If an instrument is found reading low and you have no as-found value, you cannot determine the magnitude of the error, cannot scope which product was affected, and are forced either to hold everything since the last calibration or to guess.
What happens when an instrument is out of tolerance?+
Every measurement it produced since the last confirmed good calibration is in question. The instrument is quarantined, the impact is assessed across that period, and product or decisions may need to be held. ISO 9001 requires the validity of previous results to be assessed, and the scope of that assessment depends entirely on knowing what the instrument measured.
How should calibration intervals be set?+
From stability history and consequence. An instrument that has been well within tolerance across ten consecutive calibrations may justify a longer interval; one that drifts each cycle needs a shorter one or replacement. A fleet on a uniform annual cycle has not analysed anything, and an out-of-tolerance rate of zero usually indicates intervals shorter than necessary.
Is manufacturer specification the right tolerance?+
Not necessarily. The relevant tolerance is what the decision requires. An instrument comfortably within manufacturer specification may still consume too much of the tolerance you are measuring against, which is what gauge R&R assesses. Conversely a wide specification may be entirely adequate for an indicative check.
Can we calibrate in-house?+
Yes, provided the reference standard is itself calibrated and traceable, the method is defined, the person is competent, and uncertainty is understood. In-house calibration against an uncalibrated reference provides no traceability at all, which is a common and consequential misunderstanding.
What does uncertainty change in practice?+
It determines whether a near-limit result passes. A reading 0.1 units inside the limit, from a calibration with 0.3 units of uncertainty, cannot be declared conforming without a stated decision rule. Where uncertainty is absent from the certificate, those decisions are being made by convention, and the convention becomes visible only during a dispute.
The agents
What the agents do with it
The record is short. What fails is the as-found field, the impact assessment nobody scoped, and the instrument still in use past its due date.
Holds the instrument register with required tolerance and interval, makes as-found mandatory before adjustment, and raises the impact assessment automatically on an out-of-tolerance outcome.
Links instruments to the batches and decisions they informed, so the impact scope is retrieved rather than reconstructed under time pressure.
Schedules calibration alongside maintenance, quarantines instruments past due rather than reporting them, and tracks drift history to inform interval review.
Covers instruments used for safety decisions, gas detectors and noise meters among them, where the same as-found logic applies to exposure records.
This template lives in KnowQuality — quality and food safety. HACCP, nonconformance, traceability, laboratory and customer complaints.
Sources
Sources
- ISO 9001:2015 clause 7.1.5, monitoring and measuring resources
- ISO/IEC 17025:2017 clauses 6.4, 6.5 and 7.6
- 21 CFR 211.68, automatic, mechanical and electronic equipment, FDA
- 21 CFR 820.72, inspection, measuring and test equipment, FDA
- AIAG Measurement Systems Analysis reference manual