ANSI/ASQ Z1.4 is the US standard for acceptance sampling by attributes: it tells you how many units to inspect from a lot, and how many defects to allow before rejecting it, based on the lot size and an agreed Acceptance Quality Limit (AQL). You find your plan with two table lookups, lot size to a sample-size code letter, then code letter plus AQL to a sample size and accept/reject numbers.
The full title is ANSI/ASQ Z1.4-2003 (R2018), Sampling Procedures and Tables for Inspection by Attributes and it descends directly from the military standard MIL-STD-105. It answers a question every receiving dock and final-inspection station faces: you cannot inspect every unit in a 10,000-piece lot, so how many do you check, and what result lets the lot pass? Z1.4 gives a defensible, statistically grounded answer instead of a guess like "check a few." This guide walks the tables step by step and the switching rules that make the system adapt to supplier performance.
What is AQL, and what is Z1.4 actually for?
The Acceptance Quality Limit is the worst process average, expressed as percent nonconforming or nonconformities per 100 units, that is considered acceptable as a long-run average when a continuing series of lots is submitted for sampling inspection. That definition matters, and it is widely misread. AQL is not a target defect rate for a single lot, and it is not a promise that an accepted lot contains no more than the AQL. It is a level tied to the sampling scheme so that lots at or better than the AQL have a high probability of acceptance, and lots much worse have a high probability of rejection. Z1.4 is built for a stream of lots from an ongoing supply relationship, which is why its switching rules reward good history and punish bad, rather than treating each lot in isolation.
Attributes sampling means each unit is judged pass or fail against a defined criterion, and the plan counts nonconforming units (or nonconformities). It is the right tool for go/no-go characteristics. When you are measuring a continuous variable and want to use the measurement itself, the companion standard ANSI/ASQ Z1.9 for variables sampling applies instead, and ongoing process control is better handled with statistical process control and control charts.
How do you use the Z1.4 tables step by step?
Finding a single-sampling plan under normal inspection takes five moves. You need two inputs to start: the lot size and the AQL agreed with the customer or set by your quality plan.
- Pick the inspection level. Z1.4 has three general levels (I, II, III) and four special levels (S-1 to S-4) for small samples. General Level II is the default. Level I gives smaller samples (less discrimination, lower cost); Level III gives larger samples (more discrimination). Use special levels only when small sample sizes are necessary and larger sampling risks are tolerable.
- Find the sample-size code letter. Take the lot size and inspection level to Table I. The intersection gives a code letter from A (smallest lots) to R (largest). For example, a lot of 3,200 units at General Level II lands on code letter K.
- Find the sample size and accept/reject numbers. Take the code letter and the AQL to Table II-A (the single-sampling normal-inspection table). The row gives the sample size (n) and, at your AQL column, the acceptance number (Ac) and rejection number (Re). Code letter K gives a sample size of 125; at an AQL of 1.0, that row reads Ac = 3, Re = 4.
- Draw and inspect the sample. Randomly select n units from the lot and inspect each against the criterion. Count the nonconforming units.
- Accept or reject. If the count is at or below Ac, accept the lot. If it reaches Re, reject it. In the example, 3 or fewer nonconforming units accept the lot of 3,200; 4 or more reject it. There is no in-between: Ac and Re in single sampling are consecutive numbers.
What are single, double, and multiple sampling plans?
Z1.4 offers three plan types that reach the same protection with different amounts of inspection. Single sampling inspects one sample of n units and decides. Double sampling inspects a first, smaller sample; if the result is clearly good it accepts, clearly bad it rejects, and only in the ambiguous middle does it draw a second sample and decide on the combined count. Multiple sampling extends that idea to up to seven successively drawn samples. Double and multiple plans often reduce the average number of units inspected, especially when lots are usually good or usually bad, at the cost of a more complex procedure on the floor. Many plants standardize on single sampling for simplicity and switch to double or multiple only where inspection cost is high.
| Plan type | Samples drawn | Trade-off |
|---|---|---|
| Single | One sample of n | Simplest to run; fixed inspection quantity |
| Double | One or two samples | Lower average inspection; a second sample only in ambiguous cases |
| Multiple | Up to seven samples | Lowest average inspection; most complex to administer |
How do the switching rules work?
The switching rules are what make Z1.4 a system rather than a single table, and they are the part most plants ignore and then wonder why the standard did not protect them. Inspection starts at Normal. Based on the running results, it tightens when quality slips and loosens when quality is consistently good, so the amount of scrutiny tracks the supplier's actual performance over a stream of lots.
- Normal to Tightened: switch when 2 out of 5 (or fewer) consecutive lots are rejected on original inspection. Tightened inspection uses Table II-B, which has stricter acceptance numbers.
- Tightened to Normal: switch back when 5 consecutive lots are accepted on original tightened inspection.
- Normal to Reduced: switch when the preceding 10 lots have been accepted, the total nonconforming stays under the standard's limit numbers, production is steady, and reduced inspection is approved by the responsible authority. Reduced inspection uses Table II-C and smaller samples.
- Reduced to Normal: switch back if a lot is rejected, if a lot falls in the reduced plan's gray zone between its acceptance and rejection numbers, if production becomes irregular, or if other conditions warrant it.
- Discontinue inspection: if tightened inspection is in effect and 5 consecutive lots remain on tightened without switching back, stop accepting under the standard until the supplier improves the process.
The discipline here is what actually reduces cost and risk over time. A good supplier earns reduced inspection and less cost; a slipping one triggers tightened inspection automatically, before a run of bad lots gets through. Skipping the rules and running Normal forever throws away half of what the standard is for.
Where does Z1.4 fit against SPC and process capability?
Acceptance sampling and process control answer different questions, and mature quality programs use both. Z1.4 decides whether to accept a lot that already exists, useful at receiving, at final inspection, and for a first article inspection gate. It is a sorting decision after the fact. SPC and process capability work upstream to keep the process from making defects in the first place, which is cheaper than catching them. The modern quality view treats sampling as a safety net and process control as the real prevention: if a supplier's process is capable and in control, acceptance sampling rarely rejects anything, and you can justify moving that supplier to reduced inspection or skip-lot arrangements. Where sampling keeps rejecting lots, the answer is not more inspection but fixing the supplier's process, the province of supplier quality management and incoming inspection strategy.
What do the standards say?
Sources for attributes sampling
- ANSI/ASQ Z1.4-2003 (R2018), Sampling Procedures and Tables for Inspection by Attributes is the current US attributes acceptance-sampling standard, maintained through ASQ with its variables companion ANSI/ASQ Z1.9.
- Z1.4 descends from the military standard MIL-STD-105 (the last version, MIL-STD-105E, was cancelled in 1995), and it is technically aligned with the international standard ISO 2859-1 for attributes sampling.
- The standard provides normal, tightened, and reduced plans applied with switching rules to a continuing stream of lots at a specified AQL, using sample-size code letters A through R.
- Acceptance sampling is one tool within a wider quality system; the American Society for Quality documents it alongside prevention-focused methods such as control charts (ASQ, Acceptance Sampling).
How does a connected floor change acceptance sampling?
Switching rules only work if someone is tracking the running history of lots, accepted, rejected, and how many in a row, and in many plants that history lives in a binder nobody totals until an audit. When inspection results are captured digitally at the dock and the station, the switching status updates itself: the system knows when a supplier has earned reduced inspection or tripped into tightened, so the right plan gets applied without a manual tally. That same record turns repeated rejections into a supplier-quality signal instead of a shrug, tying acceptance sampling back to the cost of quality it is meant to control. Plants that moved from paper logs to real-time inspection data, like the processor in our CLS case study can see which suppliers and defect codes actually drive rejections. See how Harmony captures floor and inspection data without a rip-and-replace on our features overview.