An out-of-tolerance (OOT) investigation is the structured process you follow when a measuring instrument is found, at calibration, to be outside its acceptance tolerance in its as-found condition. Its purpose is to determine whether any product or measurement the instrument accepted since its last in-tolerance calibration may have been accepted in error, and then to contain and disposition that impact. Both ISO/IEC 17025:2017 (Clause 7.10, nonconforming work) and AS9100 Rev D (Clause 7.1.5) require this evaluation of the validity of previous results; an as-found OOT condition closed without an impact assessment is one of the most common findings in either audit. The investigation is triggered by the as-found reading — the instrument's state before any adjustment — because that is the state in which it was making real measurements.
A defensible investigation follows a consistent sequence. First, record the as-found condition and the magnitude of the deviation before any adjustment is made; adjusting first destroys the evidence. Second, quarantine or clearly flag the instrument to prevent further use. Third, establish the affected period, which runs back to the last calibration at which the instrument was in tolerance. Fourth, perform reverse traceability — compile every measurement, lot, or product the instrument accepted during that period. Fifth, assess impact by comparing the deviation magnitude against the product tolerances and the test uncertainty ratio or guard band that applied. Sixth, determine disposition: accept as-is with documented justification, re-inspect, rework, scrap, or recall delivered product. Seventh, determine root cause — drift, mechanical damage, misuse, environmental exposure, or an interval set too long. Eighth, take corrective action, which may include repair, adjustment, replacement, or shortening the interval. Ninth, notify any affected customers when delivered product is implicated. Tenth, document the investigation and retain the records.
Reverse traceability is the heart of the investigation: it answers what the bad instrument touched. But an out-of-tolerance finding does not automatically mean product was bad. The deciding question is whether the deviation was large enough, relative to the product's tolerance, to have changed an accept/reject decision. If the instrument's error consumed only a small fraction of the product tolerance — a healthy test uncertainty ratio (TUR) with adequate guard band — previously accepted product is likely still conforming, and a documented use-as-is disposition is defensible. If the deviation approached or exceeded the guard band, the risk is real and re-inspection or recall is warranted. Quantifying this comparison, rather than reacting conservatively to every OOT, is what separates a mature program from one that triggers unnecessary recalls.
ISO/IEC 17025:2017 addresses OOT under Clause 7.10 (nonconforming work) and Clause 7.8 (reporting), requiring action when work does not conform and notification to the customer when results are affected. AS9100 Rev D Clause 7.1.5 requires evaluating and recording the validity of previous results and acting on the equipment and any affected product. For medical-device and pharmaceutical environments, FDA 21 CFR Part 820 and Part 211 impose parallel expectations for control of inspection equipment and for investigating its impact. Across all of them the common thread is identical: evaluate the impact, document it, and act.
Retain the as-found and as-left data with the deviation magnitude; the affected period and the basis for it; the list of products or measurements identified through reverse traceability; the impact assessment with the disposition and its rationale; the root-cause analysis; the corrective action taken; any customer notifications issued; and the approvals or electronic signatures that closed the investigation. These records demonstrate to an auditor that the validity of previous results was genuinely evaluated rather than assumed.
CalibrationOS includes a guided twelve-step out-of-tolerance investigation workflow. When a calibration is recorded out of tolerance, the platform captures the as-found and as-left data with uncertainty, then performs reverse traceability automatically across its calibration and usage data model to identify every product the instrument accepted in the affected period. It routes the investigation for impact assessment, root cause, and disposition with electronic signatures, and writes the entire history to a SHA-256 hash-chained, tamper-evident audit trail — producing the defensible record that ISO 17025 and AS9100 auditors expect.
It is the structured process triggered when an instrument is found outside its acceptance tolerance at calibration, in its as-found condition. The investigation determines whether measurements or product the instrument accepted since its last good calibration may have been wrongly accepted, then contains and dispositions the impact.
ISO/IEC 17025:2017 (Clauses 7.10 and 7.8) and AS9100 Rev D (Clause 7.1.5) both require evaluating the validity of previous results when measuring equipment is found out of tolerance. FDA 21 CFR Part 820 and Part 211 impose equivalent expectations for medical-device and pharmaceutical manufacturers.
The affected period runs back to the most recent calibration at which the instrument was confirmed in tolerance. Every measurement or product accepted between that point and the out-of-tolerance discovery is within scope of the impact assessment.
Reverse traceability is identifying every product, lot, or measurement that a nonconforming instrument accepted during the affected period, so the impact of its error can be assessed and the affected items contained or dispositioned.
No. Whether previously accepted product is at risk depends on the magnitude of the deviation relative to the product tolerance and the guard band that applied. A small deviation with an adequate test uncertainty ratio often supports a documented use-as-is disposition, while a deviation near or beyond the guard band warrants re-inspection or recall.
This article is licensed CC BY-SA 4.0. Share, adapt, and reuse with attribution to calibrationos.com/learn/out-of-tolerance-investigation-procedure.
Get expert insights on compliance, uncertainty, and measurement best practices.
Start managing calibrations in minutes. Free plan with 25 assets — no credit card.
Get Started Free