AQL Sampling Plan
Acceptable Quality Limit sampling plan · AQL · acceptable quality limit · ANSI/ASQ Z1.4 · ISO 2859-1 · sampling plan
A statistical rule that sets how many units to inspect from a lot and how many defects are allowed before the lot is rejected — defined by ANSI/ASQ Z1.4 and ISO 2859-1.
An AQL sampling plan answers a question every inspection department has to answer defensibly: how many units do we check, and at what point do we reject the lot? The Acceptable Quality Limit is the worst tolerable process average — expressed as a percent defective — at which a lot is still routinely accepted. The two standards in general use, ANSI/ASQ Z1.4 and its international twin ISO 2859-1, turn lot size and inspection level into a sample size code letter, then into a sample size with an accept number (Ac) and reject number (Re).
The mechanics matter. General inspection level II is the default; levels I and III trade discrimination against cost, and the special levels (S-1 to S-4) exist for destructive or expensive tests. AQLs are chosen by defect class, not by product: typically 0.65% or lower for critical, 1.0–2.5% for major, 4.0% for minor. Crucially, an AQL is not a target defect rate and it is not a specification — accepting a lot under AQL sampling is a statement about the sampling risk you have chosen, not a guarantee the lot is defect-free.
The switching rules are the part most often skipped and most often cited. A plan starts on normal inspection; two rejections in five consecutive lots switch to tightened; tightened for five consecutive accepted lots switches back; ten lots on tightened without returning means inspection stops until the supplier fixes the process. Reduced inspection and skip-lot require qualifying conditions that must be evidenced, not assumed. Running everything on normal inspection forever, or applying reduced inspection because it is cheaper, is the standard finding.
Regulated sites layer requirements on top. 21 CFR 211.84 requires representative samples of each shipment of each lot; ISO 13485 clause 8.2.6 and 21 CFR 820.250 require statistical techniques with a documented rationale for their adequacy. In V5 Ultimate the plan is master data, not an operator decision: lot size drives the sample count at the receiving or in-process step, the accept/reject number is enforced, switching state is tracked per supplier and part, and a rejection routes straight into nonconformance and supplier CAPA. See quality control software for manufacturing.
- ANSI/ASQ Z1.4
- ISO 2859-1
- 21 CFR 211.84
- 21 CFR 820.250
- ISO 13485 cl. 8.2.6
Free trial, no credit card, onboard in days, not months.
