One Result Outside Specification Is a Signal, Not a Diagnosis
An out-of-specification result starts a documented investigation. It does not, by itself, prove whether the cause was the material, the sampling process, the method, the instrument or a calculation—and a later passing result does not automatically erase it.
When a laboratory result falls outside a predefined acceptance limit, the most useful next question is not “Did the next test pass?” It is “What evidence explains the original result?” That distinction matters whenever a laboratory, procurement team or research group reviews analytical documentation.
What “out of specification” actually means
An out-of-specification, or OOS, result is a test result outside an established specification or acceptance criterion. A specification might cover identity, assay, purity, water content, microbial limits or another defined attribute. The limit and the method should be established before the sample is tested.
OOS describes the relationship between one reported result and one acceptance limit. It does not identify the cause. The result may reflect the material itself, a nonrepresentative sample, a preparation error, an instrument problem, a calculation error or another documented factor. Determining which explanation is supported requires an investigation.
The first review should protect the evidence
FDA’s OOS guidance recommends a prompt, objective laboratory assessment. That review can include the original raw data, chromatograms or spectra, calculations, instrument performance, system suitability, reference standards, solvents and reagents, and whether the approved method was followed. Where practical, original preparations should be retained long enough to test a specific hypothesis about what went wrong.
A vague label such as “analyst error” is not a complete explanation. A credible laboratory-error conclusion should identify the error, show the evidence supporting it, explain its effect on the result and preserve the original record.
Why a passing retest is not an eraser
Retesting can be part of an investigation, but its purpose should be defined in advance. For example, a retest might investigate a suspected instrument malfunction or sample-preparation problem. Repeating a test until a passing number appears is different: FDA describes that practice as “testing into compliance” and calls it unscientific.
If a documented laboratory or calculation error explains the first result, the investigation may justify invalidating it. If no such error is found, a later passing result does not provide a scientific basis for discarding the original OOS result. Both the passing and suspect results remain relevant to the evaluation.
| Evidence pattern | What it supports | What to ask for |
|---|---|---|
| A specific, documented laboratory error | A reason to evaluate whether the original result is invalid | Raw data, error description, impact assessment, approvals and corrective action |
| No error found, followed by one passing retest | Additional data—not automatic cancellation of the first result | Predefined retest plan, all results and the scientific rationale for the conclusion |
| Wide differences among portions of a sample | A possible sampling, homogeneity or preparation question | Sampling plan, sample identity, preparation records and investigation of variability |
| Similar unexpected results across lots or time | A possible trend that deserves broader review | Historical results, related lots, instruments, methods and documented follow-up |
Retesting and resampling answer different questions
Retesting
Retesting generally means analyzing another portion of the same original, homogeneous sample material. It can help test a focused hypothesis about the analytical process.
Resampling
Resampling means analyzing additional units collected under the original sampling plan or collecting a new sample from the batch when justified. It changes the sampling evidence and therefore should not be treated as interchangeable with retesting. If the original sampling method was inadequate, that problem should be identified and corrected rather than hidden by a new result.
A practical procurement checklist
When evaluating research products for sale or reviewing documentation from a supplier, an OOS history is more useful when the record preserves the full decision trail. Ask whether the documentation shows:
- the exact product, lot, sample and test method;
- the specification and acceptance limit in effect when testing began;
- the original result and underlying raw data;
- whether system suitability and instrument checks passed;
- the investigation hypothesis and the evidence used to test it;
- the number and purpose of any retests or resamples;
- all results—not only the final passing value;
- the conclusion, approval, corrective action and any review of related lots.
A certificate of analysis is a summary, not the entire laboratory record. When a report contains an unusual result, amended value or retest note, the lot number and method should stay connected to the supporting investigation. A clean-looking final number is less informative than a transparent record of how the conclusion was reached.
The simplest rule: preserve context
Good analytical review does not treat every OOS result as proof of defective material, nor every passing retest as proof that nothing happened. It keeps the first result, the investigation and the conclusion connected. That approach makes documentation more auditable and helps research teams distinguish evidence from assumption.
Primary sources
- U.S. Food and Drug Administration, Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production (October 2006; accessed October 16, 2026).
- Electronic Code of Federal Regulations, 21 CFR 211.192, Production record review (page current through October 14, 2026; accessed October 16, 2026).
Keep the evidence connected to the lot
Use the bwell journal for practical research-literacy guides, and review available certificates by matching the product name and lot rather than assuming one report applies to every item.
Editorial method: Published by bwell. AI tools assisted with organization and copyediting; the article was checked against the linked primary sources and the current bwell site. This educational content does not provide medical, clinical or legal advice.