What AQL Actually Means
Acceptable Quality Limit (AQL) is the maximum percentage of defective units in a lot that is considered acceptable for the purpose of sampling inspection. It is NOT the acceptable defect rate in your product — it is a statistical tool for making lot disposition decisions based on a sample.
This distinction matters because it is frequently misunderstood. An AQL of 0.65% for major defects does not mean you accept 0.65% defective units in every lot. It means your sampling plan is designed so that lots with a true defect rate at or below 0.65% will be accepted approximately 95% of the time. Lots with higher defect rates face increasing probability of rejection. — [ANSI/ASQ Z1.4-2003 (R2018)]
AQL sampling is conducted by the Quality Unit — not production operators — and must NOT be performed by the same operator who conducted the 100% inspection. Samples must be selected randomly across the lot. — [USP <1790>, 2017; 21 CFR 211.165]
How AQL Sampling Works in Visual Inspection
After 100% visual inspection of a lot, AQL sampling provides a statistical verification that the inspection process worked effectively:
- Determine lot size: Total units in the inspected lot
- Select inspection level: Level II is the default standard; Level I for less stringent, Level III for more rigorous — [ANSI/ASQ Z1.4, Table I]
- Find sample size: Use the lot size and inspection level to determine the code letter, then find the corresponding sample size
- Draw random sample: Select units randomly from across the lot (not just from the top or edges)
- Inspect and classify: Examine each sample unit and classify any defects as critical, major, or minor
- Apply accept/reject criteria: Compare defect counts against the acceptance number (Ac) and rejection number (Re) for each defect category
Typical AQL Values for Visual Inspection
| Defect Category | Normal AQL | Tightened AQL | Rationale |
|---|---|---|---|
| Critical | 0.01–0.1% | 0.01–0.065% | Patient safety risk; sterility compromise |
| Major | 0.4–0.65% | 0.25–0.4% | Reduces product usability |
| Minor | 2.5–4.0% | 1.5–2.5% | Cosmetic only; no safety impact |
These values are industry-typical but must be justified for your specific products and risk profile. Your defect classification system defines which defects fall into each category. — [USP <1790>, 2017; ISO 2859-1:1999]
Switching Rules
ANSI/ASQ Z1.4 defines mandatory switching rules that adjust inspection stringency based on lot history:
- Normal → Tightened: When 2 out of 5 consecutive lots fail AQL under normal inspection. This is mandatory — you cannot remain on normal inspection when quality is deteriorating.
- Tightened → Normal: When 5 consecutive lots pass under tightened inspection. Quality has demonstrated recovery.
- Normal → Reduced: When 10 consecutive lots pass under normal inspection AND production is stable. Allows smaller sample sizes.
- Reduced → Normal: When a lot fails under reduced inspection. Immediate return to standard sampling.
These switching rules must be implemented in your procedures and followed without exception. Failure to apply switching rules is a compliance gap that auditors will identify. — [ANSI/ASQ Z1.4-2003, Section 9]
What to Do When a Lot Fails
An AQL failure triggers mandatory actions per FDA expectations:
- Quarantine the lot: Do not release for distribution
- Investigate: Determine root cause — was it an inspection failure, a production issue, or a sampling anomaly? — [21 CFR 211.192]
- Reinspection decision: Reinspection (100% re-examination) is allowed only with documented justification and tightened acceptance limits. The FDA Draft Guidance states that tertiary re-inspection is NOT typically justifiable. — [FDA Draft Guidance, 2021]
- CAPA: Implement corrective and preventive actions based on root cause
- Trending: Record the failure and evaluate against lot history for switching rule application
The FDA Draft Guidance (2021) explicitly states that investigations are mandatory when AQL failures occur. Simply reinspecting and releasing without understanding why the lot failed is not acceptable. — [FDA Draft Guidance, 2021]
Understanding Reject Rates
Reject rates from 100% inspection are distinct from AQL results but equally important for process understanding:
- Overall reject rate: Total units rejected divided by total units inspected — indicates your process cleanliness and inspection sensitivity
- Reject rate by category: Breaking down rejects by defect type reveals specific process issues
- False reject rate: Units rejected that are actually conforming — particularly relevant for AVI systems (typically 0.5–3%)
- Trending: Track reject rates over time by product, line, shift, and operator to identify patterns
A sudden increase in reject rate may indicate a process issue (contamination event, equipment wear) or an inspection issue (overly sensitive AVI recipe, fatigued operator). Both require investigation. — [FDA Draft Guidance, 2021]
AQL for Automated Inspection
When using AVI systems, AQL sampling serves an additional purpose: verifying that the automated system is performing as validated. The AQL sample should be inspected manually by qualified inspectors to confirm the machine's accept/reject decisions are appropriate.
Key considerations for AVI + AQL:
- AQL inspectors must be qualified separately from AVI operation
- Sample must include units from throughout the production run
- Any discrepancy between AVI disposition and AQL findings requires investigation
- AVI false reject rate should be monitored separately from true defect rate
Common Mistakes
- Same operator for 100% and AQL: This violates the independence requirement — AQL must be performed by different, qualified personnel
- Non-random sampling: Taking samples only from accessible locations (top of tray, front of pallet) introduces bias
- Ignoring switching rules: Remaining on normal inspection after multiple failures is non-compliant
- AQL as the only control: AQL sampling verifies 100% inspection effectiveness — it does not replace 100% inspection
- No investigation of failures: Every AQL failure requires documented investigation per 21 CFR 211.192
References
- ANSI/ASQ Z1.4-2003 (R2018), "Sampling Procedures and Tables for Inspection by Attributes."
- ISO 2859-1:1999/Amd 1:2011, "Sampling procedures for inspection by attributes."
- USP General Chapter <1790>, "Visual Inspection of Injections," 2017.
- FDA Draft Guidance, "Inspection of Injectable Products for Visible Particulates," December 2021.
- 21 CFR 211.165, "Testing and release for distribution."
- 21 CFR 211.192, "Production record review."
