Inspecting every unit in a 10,000-piece shipment isn't just slow — it's often less reliable than a properly designed sample. How lot size, inspection level, and AQL value combine into a sample size, and what an accepted lot doesn't guarantee.
Inspecting every single unit in a 10,000-piece
shipment isn’t just slow — for many defect types it’s also less
reliable than a properly designed sample, since inspector fatigue sets
in long before unit 10,000. AQL sampling plans exist to make “how many
do we actually need to check” a statistical decision instead of a guess.
Acceptable Quality Level (AQL) is the maximum percent defective that,
for sampling purposes, can be considered acceptable as a process
average. ANSI/ASQ Z1.4 (based on the older MIL-STD-105) is the most
common sampling standard used to translate a lot size and an AQL value
into an actual sample size and an accept/reject number.
| Input | What it determines |
|---|---|
| Lot size | Total number of units in the batch being inspected |
| Inspection level | How aggressively sample size scales with lot size (general levels I, II, III) |
| AQL value | The acceptable percent defective, which sets the accept/reject thresholds |
Z1.4 isn’t a static lookup — it includes switching rules that move a
supplier between normal, tightened, and reduced inspection based on
recent lot history. A pattern of rejected lots triggers tightened
inspection with stricter accept numbers, while a strong track record
can qualify a supplier for reduced sample sizes, which is the incentive
structure that makes the standard self-correcting over time.
An accepted lot under an AQL plan can still contain defective units
— the standard controls the statistical risk of accepting a bad lot, it
doesn’t promise zero defects got through. For characteristics where
even a single failure is unacceptable, such as a safety-critical
dimension, AQL sampling generally isn’t the appropriate control; 100%
inspection or a control-chart-based approach fits better.
AQL sampling fits inside the broader quality systems work covered in
SCMEP’s Quality and Inspection training
catalog. As a NIST Manufacturing Extension
Partnership affiliate serving South Carolina manufacturers since
1989, our focus is helping teams choose sampling plans that fit
their actual risk tolerance, not a default table setting.
If your team is setting up incoming inspection sampling or
evaluating supplier switching rules, you can
browse the Quality and Inspection
training catalog or email the training team.
An AQL sampling plan only produces trustworthy results if the gage doing the measuring is actually in calibration. Read our guide to gage calibration intervals — setting and managing a calibration program.
AQL is the maximum percent defective that, for sampling purposes, can be considered acceptable as a process average. ANSI/ASQ Z1.4 is the most common standard used to translate a lot size and AQL value into a sample size.
An AQL plan is built from the lot size, the inspection level (which determines how sample size scales with lot size), and the AQL value, which sets the accept and reject thresholds.
Z1.4 includes switching rules that move a supplier between normal, tightened, and reduced inspection based on recent lot history — a pattern of rejected lots triggers stricter inspection, while a strong record allows reduced sampling.
No. AQL sampling controls the statistical risk of accepting a bad lot but doesn’t guarantee zero defects got through. Safety-critical characteristics generally call for 100% inspection instead.
