When a periodic calibration check shows an instrument reading outside its stated tolerance limits, that reading is an out-of-tolerance (OOT) condition, and it triggers a specific, documented response — not just a re-calibration. Under ISO/IEC 17025:2017 and the GxP frameworks that reference it, an out-of-tolerance finding is treated as a quality event: the lab must quarantine the instrument, assess whether any prior results are affected, document the finding, and record a corrective action before the instrument goes back into service. This guide walks through that procedure end to end.
Out of tolerance vs. out of calibration
These two phrases are often used loosely as if they mean the same thing, but they describe different states:
- Out of tolerance (OOT) means the instrument was checked against a reference standard and the deviation exceeded the tolerance limit defined in the SOP, method, or manufacturer specification. The instrument may still be functioning — it is simply reading outside the accuracy band the lab requires for that application.
- Out of calibration is a broader, less precise term generally used to describe an instrument whose calibration has lapsed (the due date passed) or that is not currently calibrated at all. An instrument can be technically “in calibration” (current cal sticker, not yet due) and still be found out of tolerance at its next check — that is exactly the scenario this guide covers.
In practice, an OOT finding usually surfaces during a scheduled recalibration or a mid-interval verification check, when the as-found readings are compared against tolerance before any adjustment is made.
Step 1: Quarantine the instrument immediately
The moment an as-found check shows a deviation outside tolerance, remove the instrument from service. ISO/IEC 17025:2017 Clause 6.4 requires that equipment shown to be outside specified requirements be isolated or clearly labeled to prevent inadvertent use until the situation is resolved. In practice this means:
- Attach an “OUT OF SERVICE – DO NOT USE” tag or label, separate from the routine calibration-due sticker.
- Physically remove or lock out the instrument if it is used by multiple analysts or shared across a core facility.
- Log the OOT event in the equipment record or LIMS/CMMS the moment it is found, before the impact assessment is complete — the finding itself, not just the resolution, needs a timestamp.
Step 2: Assess the impact on prior results
This is the step most often shortchanged, and the one accreditation assessors and quality auditors focus on. An OOT reading doesn’t just affect the next test run — it calls into question every result generated on that instrument since the last time it was confirmed to be in tolerance. ISO/IEC 17025:2017 Clause 7.10 (nonconforming work) requires the laboratory to evaluate the significance of the nonconformity, including an impact analysis on previously issued results.
A defensible impact assessment typically covers:
- Look-back period: from the date/time of the last successful in-tolerance check to the date/time the OOT was found.
- Affected work: every sample, batch, run, or test result produced on that instrument during the look-back period — pulled from the LIMS, ELN, or logbook, not from memory.
- Magnitude of deviation: how far outside tolerance the instrument was found, and in which direction — a balance reading 0.3% high has a different practical effect than one reading 8% high.
- Effect on decisions: whether any pass/fail determination, release decision, or reported result would change if corrected for the actual deviation. This is the determination that decides whether results need to be recalled, reissued, or flagged, or whether the deviation was immaterial to the outcome.
Document the impact assessment even when the conclusion is “no effect” — a documented negative finding is what an assessor or auditor expects to see; an absence of any assessment reads as a gap regardless of the outcome.
Step 3: Investigate the root cause
Before returning the instrument to service, determine why it drifted out of tolerance. Common root causes include:
- Normal drift that exceeded the calibration interval set for that instrument — a signal the interval itself may be too long.
- Physical damage, mishandling, or an environmental excursion (temperature, humidity, vibration, power interruption).
- A worn or contaminated component (e.g., a pH electrode nearing end of life, a pipette seal, a balance pan).
- An error in the prior calibration itself, or use of an out-of-date reference standard.
The root cause determines the corrective action: a recalibration and adjustment is not the same fix as replacing a worn part, shortening the calibration interval, or retraining an operator on handling procedure.
Step 4: Document on an out-of-tolerance form / nonconformance record
Most labs use a dedicated out-of-tolerance calibration form, or fold the OOT into the lab’s general nonconformance report (NCR) process — either is acceptable under ISO/IEC 17025 as long as the required elements are captured. At minimum, the record should include:
- Instrument ID, asset number, and location
- Date/time the OOT was discovered, and by whom
- As-found readings vs. tolerance limits, with the reference standard used
- Date/time of the last known in-tolerance check (the start of the look-back period)
- Impact assessment findings and conclusion (Step 2)
- Root cause (Step 3)
- Corrective action taken, and who approved returning the instrument to service
- As-left readings after adjustment/repair, confirming the instrument is back within tolerance
- Notification record, if any affected result had to be communicated to a client, sponsor, or regulatory body
This record is what an accreditation body, FDA investigator, or internal quality audit will ask to see first if they identify an instrument that had an OOT event on file — treat it as the artifact of record, not a formality to file away.
Step 5: Notify affected parties, where required
Whether notification is required depends on the regulatory context and what the impact assessment found:
- ISO/IEC 17025-accredited labs: if the nonconformity affected the validity of results already reported to a client, the lab is required to notify the client and, where relevant, recall or correct the affected reports.
- GxP-regulated environments (GLP/GMP/GCP, per GxP compliance requirements): an OOT on qualified/validated equipment typically triggers a deviation report, and may require notifying quality assurance and, depending on scope, a regulatory filing if it affects a released batch or a reported study result.
- CLIA-certified clinical labs: an OOT affecting patient testing generally requires corrected-report procedures if previously reported patient results are shown to be affected, consistent with the lab’s CAP/CLIA quality management requirements.
Even when formal notification isn’t triggered, internal escalation to the quality manager or technical director is standard practice the moment an impact assessment identifies any affected result.
Step 6: Return the instrument to service
Before the “OUT OF SERVICE” tag comes off, confirm and document:
- The instrument has been recalibrated, adjusted, or repaired and re-verified as-left within tolerance, ideally by an accredited calibration provider that issues a new certificate (see calibration certificates and metrological traceability).
- The corrective action from Step 3 has actually been implemented, not just proposed.
- A responsible person (quality manager, lab director, or designated approver per the SOP) signs off on returning the instrument to service.
- The equipment record, LIMS asset record, and any physical calibration-due label are updated with the new calibration date and next-due date.
Preventing repeat out-of-tolerance findings
A single OOT event is a normal, expected part of instrument life. Repeated OOT findings on the same instrument are a signal to act, not just to keep documenting:
- Shorten the calibration interval for that specific instrument rather than assuming the standard interval still applies.
- Add or tighten an interim (mid-interval) verification check between full calibrations, so drift is caught before it accumulates.
- Review handling, environmental controls, and operator technique — especially for instruments sensitive to technique, such as pipettes, pH meters, and analytical balances.
- Consider whether the tolerance itself, not the instrument, is the problem — an overly tight tolerance relative to the instrument’s actual capability will generate chronic, low-value OOT events.
Frequently asked questions
Is an out-of-tolerance finding the same as a failed calibration?
Not exactly. A calibration itself (the act of comparing the instrument to a reference standard and recording the readings) doesn’t “fail” — it produces data. That data either shows the instrument within tolerance or outside it. “Out of tolerance” describes the outcome of the comparison, not a failure of the calibration process itself.
Does every out-of-tolerance instrument need a full recall of prior results?
No. A recall or correction of previously reported results is only required when the documented impact assessment (Step 2) concludes that the deviation was large enough, and the affected results close enough to a decision threshold, that the reported outcome would plausibly have been different. Many OOT findings are minor deviations with no material effect on any reported result — but that conclusion still has to be reached and documented, not assumed.
Who is responsible for deciding whether an OOT is significant?
Under ISO/IEC 17025, this is a quality management function — typically the quality manager or technical manager, following the lab’s documented nonconformance procedure, not the analyst who happened to discover the OOT reading.
How long should out-of-tolerance records be retained?
Retention should follow the lab’s accreditation body and regulatory requirements for quality records generally (commonly aligned with sample/result retention periods), not a shorter, informal retention period specific to calibration paperwork.
For the broader calibration program this procedure sits inside, see calibration certificates and metrological traceability, what ISO/IEC 17025 actually accredits, and nonconformity: major vs. minor and the NCR/NCAR process. For instrument-specific calibration procedures, see the guides on pipette calibration, analytical balance calibration, pH meter calibration, and spectrophotometer calibration.
