What this is
What is an out of specification investigation?
What is an out of specification investigation?
An out of specification investigation is the structured review that follows any test result falling outside its approved specification. It exists to decide whether the result is genuine before anyone acts on it, running first against the laboratory itself and only then against the process that produced the sample.
When is an out of specification investigation raised?
It is raised the moment a result comes back outside specification, not reconstructed afterwards once someone has already decided what happened. The trigger is the result, not the eventual product decision.
What is the difference between confirming and invalidating an OOS result?
Confirming a result means the investigation found nothing wrong with the laboratory work or the process, so the out of specification value stands and the product decision follows it. Invalidating a result means a documented, independently approved cause was found — a transcription error, an equipment fault, a sampling mistake — that means the number does not represent the sample. Invalidating without that evidence is a third, unacceptable option the template records separately so it can be found.
Scope
When is an out of specification investigation required?
This record is one step in a larger programme. Using it for work that belongs to a neighbouring template produces records that are hard to report on later.
Use this template when
- A test result has come back outside its approved specification and needs a documented decision before anyone acts on it
- The workspace is being set up, or the register needs an entry added or retired
- You are running the Laboratory and Testing programme and this is one of its steps
- A linked record needs this one to exist: links Micro results, Holds, CAPA
- A retest is being considered and the protocol governing it needs to be agreed and dated before testing starts
Do not use it for
- Laboratory Sample Log, which logs every sample taken for testing, with what it is, why it was taken and where it went.
- Microbiological Result Record, which records a micro result against its specification, with the action taken where it is out.
- Chemical and Compositional Result Record, which records analytical results for composition, additives, contaminants or nutritional claims.
- Laboratory Method Verification, which confirms a method performs as claimed before it is used to generate reportable results, not after a result is already in dispute.
- Anything outside KnowQuality, which belongs in the workspace that owns that process
Compliance mapping
Which ISO 17025 cl.7.10 requirements does this satisfy?
ISO 17025 cl.7.10 requires a laboratory to control nonconforming work, including deciding whether a result is valid before it is acted on. The clause map below shows where each investigation step earns that clause its evidence.
| Clause | Requirement | Where it lands |
|---|---|---|
| ISO 17025 cl.7.10 | Nonconforming work must be identified and controlled before it affects a result-based decision. | Header |
| ISO 17025 cl.7.10 | The laboratory's own performance — analyst, method, equipment, calculation, original data — is checked before any process cause is entertained. | Phase one, the laboratory |
| ISO 17025 cl.7.10 | Any retest must follow a protocol agreed and documented before testing, not improvised once a result is known. | Retesting |
| ISO 17025 cl.7.10 | Where laboratory causes are ruled out, the investigation extends to the process and sample that generated the result. | Phase two, the process |
| ISO 17025 cl.7.10 | The significance of the nonconforming work must be evaluated and the decision recorded with its basis. | Conclusion |
| ISO 17025 cl.7.10 | Invalidating a result requires independent authorisation, not the reporting analyst's own sign-off. | Conclusion |
| ISO 17025 cl.7.10 | Corrective action is raised where the investigation identifies a cause requiring one, and the record closes with that evidence attached. | Result |
What it does not cover
- Retest Protocol Agreed Before Testing, which is marked "After the fact" or "None" — the retest happened first and the rule got written to fit it.
- Invalidation Approved Independently, which is marked "No" — the analyst who produced the inconvenient result is also the one who made it disappear.
- Analyst Interviewed, which is marked "No" — the investigation proceeded without asking the person who ran the test what happened.
- Original Data Reviewed, which is marked "No" — nobody went back to the raw data before deciding the result was wrong.
- Root Cause Identified, which is marked "No" while the result is recorded as invalidated — there is no accepted explanation for why the number is being thrown out.
Global
Out of Specification Investigation requirements by country
An out of specification investigation gets read differently depending on who is asking for it, and the standard it has to survive is rarely local law.
FDA Guidance for Industry, Investigating Out-of-Specification Test Results for Pharmaceutical Production
Retesting into compliance is named directly and treated as unacceptable without a documented, pre-agreed investigation and protocol.
A site supplying into the US should expect an OOS record without a dated retest protocol to be treated as a data integrity finding, not a paperwork gap.
UKAS accreditation under ISO/IEC 17025
Accreditation bodies audit nonconforming work control as a condition of keeping the scope of accreditation, not as good-practice advice.
A UK laboratory that cannot produce the independent invalidation signature on demand risks a finding against its accredited scope, not just the individual result.
ILAC-G8 guidance on decision rules and out of specification results
Sets out how a laboratory should decide and document whether a result is genuinely nonconforming before acting on it, independent of any single national regulator.
A laboratory issuing results across borders should hold its OOS process to this reference regardless of local law, since customers increasingly ask for it by name.
How to complete it
How to complete an out of specification investigation, step by step
Filling in every field does not make an OOS investigation defensible. Four judgement calls decide whether it survives scrutiny months later.
An investigation that jumps to a process deviation without a clean answer on analyst, method, equipment and calculation is vulnerable — auditors read that sequence as evidence the laboratory cause was never seriously tested, whatever phase two eventually concludes.
The date on Retest Protocol Agreed Before Testing has to sit before the date on the retest itself. If the paperwork cannot establish that ordering, a clean retest result does not rescue the investigation — the sequence is the defence, not the number.
Confirmed and invalidated with evidence both rest on a documented basis. Invalidated without evidence has none, and a conclusion resting on that option is the fastest way to turn an OOS investigation into a data integrity citation rather than a closed record.
Independent means someone who did not run the original test and has no stake in the product decision. A second signature from the same analyst's own supervisor, who also signed off the original batch, is independent in name only.
What auditors find
Most common out of specification investigation findings
Six patterns turn up repeatedly when OOS investigations are pulled for audit, and most trace back to sequencing rather than missing fields.
| Finding | Clause | What fixes it |
|---|---|---|
| Retest protocol dated the same day as the retest result, or not dated at all. | ISO 17025 cl.7.10 | Require the protocol to carry its own timestamp and signature, entered and locked before any retest sample is logged against this record. |
| Phase two conclusions reached without a completed phase one. | ISO 17025 cl.7.10 | Make the process section conditional on Analyst Interviewed, Method Followed Correctly and Original Data Reviewed all being answered first. |
| Result invalidated with only the original analyst's own signature attached. | ISO 17025 cl.7.10 | Route the Second Signature field to a role that never held the original sample, enforced at the workspace level rather than left to habit. |
| Trend Supports The Result recorded as uncertain when other batches were never actually pulled and reviewed. | ISO 17025 cl.7.10 | Tie Other Batches Reviewed to Trend Supports The Result so uncertain cannot be selected without a record of what was checked. |
| Investigation Required marked No on a result that was later escalated anyway once the retest failed. | ISO 17025 cl.7.10 | Score Investigation Required on the result's potential product impact at the time of the OOS, not on what the eventual retest or RCA showed. |
| CAPA ID left blank against an Action Required of Yes for weeks after the investigation closes. | ISO 17025 cl.7.10 | Block the record from showing Complete while Action Required is Yes and CAPA ID is still empty. |
Case in point
Case in point: the retest that arrived a week after the protocol
A finished-product pH result comes back outside specification. The analyst reruns the same sample twice before telling anyone, gets one pass and one fail, and reports the pass. Phase one is only completed after the fact, and Retest Protocol Agreed Before Testing is answered "After the fact" because no protocol existed before the first rerun.
The investigation is reopened. This time the original sample is retained, a protocol specifying one confirmatory retest by a second analyst is signed before the retest happens, and the earlier undocumented reruns are logged as a deviation in their own right rather than folded into the result. The conclusion — invalidated with evidence — now has something to point to.
The template
The template, field by field
The form exactly as it installs. Every field, option, score and conditional rule is editable, and the links to other templates come with it.
6 sections
- Reference
- QUA-088
- Archetype
- Record
- Record ID
- OOS-2026-000
- Scoring
- Investigations closed
- Direction
- High is good
- Singleton
- Yes
- Basis
- ISO 17025 cl.7.10
- Links
- Links Micro results, Holds, CAPA
- Tags
- Laboratory, Investigation
- Sections
- 6
- Fields
- 53
- Follow up fields
- 5
- Repeating sections
- 0
- Links out
- 5
Header
10 fieldsInvestigation ID*
Auto sequence. Format OOS-2026-000.
The record's own ID. Other templates point at this value.
Status*
Drives who this goes to next.
- Planned2 pts
- In progress2 pts
- Complete3 pts
- Deferred0 pts
- Open0 pts
- Closed3 pts
- Overdue0 pts
Date and Time*
Completed By*
Site*
Site ID*
Format SITE-000.
Links to FDN-001 Site ID
Do Not Test Until You Like The Answer
Retesting without a documented reason and a protocol agreed beforehand is the single most criticised practice in laboratory investigation. Decide the rules before you know the answer.
Source Result ID*
Immediate predecessor record
Parameter*
Result Versus Limit*
Phase one, the laboratory
9 fieldsAnalyst Interviewed*
- Yes3 pts
- No0 pts
Method Followed Correctly*
- Yes3 pts
- Deviation found0 pts
Calculation Checked*
- Yes3 pts
- No0 pts
Equipment In Calibration*
- Yes3 pts
- Overdue0 pts
Reagents And Media In Date*
- Yes3 pts
- No0 pts
Controls Performed Correctly*
- Yes3 pts
- Partly1 pt
- No0 pts
Original Data Reviewed*
- Yes3 pts
- No0 pts
Transcription Error Found*
- No3 pts
- Yes0 pts
Obvious Laboratory Error Identified*
- No2 pts
- Yes0 pts
Retesting
6 fieldsRetest Protocol Agreed Before Testing*
Number of retests, who does them and what result decides the outcome, all agreed in writing first.
- Yes3 pts
- After the fact0 pts
- None0 pts
Number Of Retests Permitted
Retest Justification Documented*
- Yes3 pts
- No0 pts
Original Sample Retested
- Yes3 pts
- Not possible2 pts
- No0 pts
Reserve Sample Tested
- Yes3 pts
- None available0 pts
- No1 pt
Retest Results
Phase two, the process
6 fieldsProduction Records Reviewed*
- Yes3 pts
- Partly1 pt
- No0 pts
Process Deviation Found*
- No3 pts
- Yes0 pts
Sampling Error Considered*
- Yes3 pts
- No0 pts
Other Batches Reviewed*
- Yes3 pts
- No0 pts
Trend Supports The Result*
- Yes1 pt
- Isolated result2 pts
- Uncertain1 pt
Extent Of Condition ID
Links to FDN-018 Review ID
Conclusion
9 fieldsResult Confirmed Or Invalidated*
- Confirmed1 pt
- Invalidated with evidence3 pts
- Invalidated without evidence0 pts
Basis For The Conclusion*
Invalidation Approved Independently*
Nobody invalidates their own inconvenient result. It needs a second signature.
- Yes3 pts
- Not applicable3 pts
- No0 pts
Product Decision*
Hold ID
Links to QUA-003 Hold ID
Root Cause Identified*
- Yes3 pts
- Probable1 pt
- No0 pts
Investigation Required*
Set by potential outcome, not by what actually happened.
- No3 pts
- Yes0 pts
Investigation Level
RCA ID
Format RCA-2026-00000.
Links to FDN-013 RCA ID
Result
13 fieldsItems Assessed*
Excludes anything marked N/A.
Items Failed*
Score Percent*
Calculated on submission. High is good. N/A items leave the denominator.
Result Band*
- Pass3 pts
- Caution1 pt
- Fail0 pts
Completeness Percent*
How much of the template was actually answered. A high score on a half completed form is not a high score.
Action Required*
Raise the action record, then enter its reference here.
- No2 pts
- Yes0 pts
Priority
- High0 pts
- Medium1 pt
- Low3 pts
CAPA ID
Format CAPA-2026-00000.
Links to FDN-014 CAPA ID
Action Owner
Laboratory Lead*
Signature*
Quality Manager*
Second Signature*
QUA-088 · record IDs look like OOS-2026-000 · Links Micro results, Holds, CAPA
Open in KnowellaRun it with agents
From a document you fill in to a programme that runs itself
The form is the easy part. Keeping the protocol dated ahead of the retest, chasing the second signature and holding the evidence together is the work that actually slips.
Holds the OOS investigation against the source result and the register, flags any retest logged before its protocol is signed, and chases the independent second signature before the record can close.
Confirms the calibration status behind Equipment In Calibration at the moment the OOS was raised, so the investigation isn't relying on a memory of when the last check happened.
Checks the analyst's training record against the method in question, so Analyst Interviewed and Method Followed Correctly aren't answered from assumption.
Coordinates the crew across phase one and phase two, rolls open investigations and overdue retests into one view, and holds every write for your approval before it touches a record.
This template lives in KnowQuality — quality and food safety. HACCP, nonconformance, traceability, laboratory and customer complaints.
Meet KnowQuality→Glossary
Out of Specification Investigation definitions and key terms
- Out of specification (OOS)
- A test result that falls outside the approved specification or acceptance criteria for the parameter being measured.
- Testing into compliance
- The rejected practice of retesting a sample repeatedly, without a pre-agreed protocol, until a passing result appears, and reporting only that one.
- Phase one investigation
- The laboratory-focused stage of an OOS review — checking analyst, method, equipment and calculation — before looking beyond the lab.
- Phase two investigation
- The production-focused stage of an OOS review, opened only once laboratory error has genuinely been ruled out.
- Result invalidation
- Formally setting aside a test result as not representative of the sample, on a documented basis and with independent approval.
FAQ
Frequently asked questions about out of specification investigation
What is an out of specification investigation?+
It is a record used in KnowQuality that investigates an out of specification result to decide whether it is real before acting on it. It is built against ISO 17025 cl.7.10 and forms part of the Laboratory and Testing programme.
When is an OOS investigation raised?+
It is raised on every out of specification result, at the moment the result comes back, rather than reconstructed later once a retest has already happened.
What is the difference between phase one and phase two of an OOS investigation?+
Phase one checks the laboratory itself — the analyst, the method, the equipment, the calculation and the original data. Phase two only opens once phase one has genuinely ruled out laboratory error, and it checks the production process, other batches and whether the result fits a trend.
Who can invalidate an out of specification result?+
Nobody invalidates their own inconvenient result. The template requires an independent second signature before a result can be invalidated, and records separately whether that invalidation came with documented evidence or without it.
How many retests are allowed after an OOS result?+
There is no fixed number in the standard. The number permitted, who performs them and what result governs the outcome must be agreed and documented in a protocol before the first retest is run, not decided afterwards to fit whatever result comes back.
Which programme does the out of specification investigation belong to?+
It is part of Laboratory and Testing. That programme covers sampling, method verification, proficiency testing and this investigation, and its outcome is results you can defend, with retesting governed by a protocol written before the retest.
Keep going
Related templates and programmes
Industries this is written for
Programmes this belongs to
Used together in Laboratory and Testing
Maintenance Calibration Record
Records calibration of maintenance instruments such as torque wrenches, pressure gauges and multimeters
Calibration Record
Records the calibration of a piece of test equipment, including standard used, result and next due date
Laboratory Sample Log
Logs every sample taken for testing, with what it is, why it was taken and where it went
Microbiological Result Record
Records a micro result against its specification, with the action taken where it is out
Chemical and Compositional Result Record
Records analytical results for composition, additives, contaminants or nutritional claims
Laboratory Method Verification
Verifies that a test method performs as expected in your hands, on your matrix, before results are relied on
More in Laboratory
Laboratory Sample Log
Logs every sample taken for testing, with what it is, why it was taken and where it went
Microbiological Result Record
Records a micro result against its specification, with the action taken where it is out
Chemical and Compositional Result Record
Records analytical results for composition, additives, contaminants or nutritional claims
Laboratory Method Verification
Verifies that a test method performs as expected in your hands, on your matrix, before results are relied on
Proficiency Testing Record
Records participation in a ring trial or proficiency scheme and how the result compared with the consensus
Laboratory Environmental Check
Checks laboratory conditions that affect results, covering temperature, humidity, cleanliness, segregation and media storage
Written and reviewed by
Siddarth Singh
Founder & Chief Executive Officer, Knowella
Certified Safety Professional and industrial and systems engineer with more than a decade inside food supply chain, freight and manufacturing operations. This page was written against the current text of the standards it cites, not against secondary summaries of them.
- Certified Safety Professional (CSP), Board of Certified Safety Professionals
- MBA, University of Chicago Booth School of Business
- MS and BS, The Ohio State University, Industrial and Systems Engineering
- Six Sigma Black Belt
Sources and last review. Reviewed 16 August 2026 against:
- ISO/IEC 17025:2017 cl.7.10 — Control of nonconforming work
- FDA Guidance for Industry — Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production
- ILAC-G8:09/2019 — Guidelines on Decision Rules and Statements of Conformity
This page is general guidance, not legal advice. Confirm requirements with your jurisdiction’s regulator.