ANSI/ASQ Z1.4 sampling plans let you decide whether to accept or reject a batch of goods by inspecting a random sample and comparing the defect count to a table-defined acceptance number. To build a plan, you need three inputs: the lot size, the inspection level, and the Acceptable Quality Limit (AQL). Those three values point you to a sample size code letter, and the code letter points you to the sample size, acceptance number, and rejection number for your inspection. The current edition is ANSI/ASQ Z1.4-2003, reaffirmed in 2018, and its international counterpart is ISO 2859-1, which uses the same table structure and switching logic.
The Three Inputs Every Plan Needs
Nothing in the tables works without an accurate lot size, a chosen inspection level, and a specified AQL. Get any of these wrong and the numbers you pull will not mean what you think they mean.
Lot Size
The lot size is the total number of units in the batch. Table I groups lot sizes into ranges, and the ranges have hard edges: a lot of 490 units can fall in a different range than a lot of 501 units and produce a different sample requirement. Count the lot before you open the standard.
Inspection Level
The inspection level controls sample size relative to lot size. Z1.4 defines three general levels (I, II, III) and four special levels (S-1 through S-4). General Level II is the default when the contract or purchase order is silent. Level I uses smaller samples and fits suppliers with a strong track record. Level III uses larger samples and improves discrimination when the stakes are higher or the supplier is unproven. The special levels S-1 through S-4 produce very small samples and exist for destructive, time-consuming, or expensive tests.
Acceptable Quality Limit
The AQL is the worst process-average defect rate the buyer is willing to tolerate. Values you will see repeatedly include 0.065, 0.10, 0.25, 0.65, 1.0, 1.5, 2.5, 4.0, and 6.5. Lower AQLs tighten the acceptance criteria; higher AQLs loosen them. The AQL belongs in the contract or purchase order. Leaving it to be decided at the factory defeats the purpose of using a documented sampling standard at all.
Defect Classification and AQL Assignment
Most programs run more than one AQL because not every defect carries the same cost. A typical scheme uses three categories:
- Critical defects: safety hazards or violations of mandatory regulations. The AQL is usually set at zero, so no critical defects are acceptable in the sample.
- Major defects: problems that make the product unusable or significantly reduce its function. A common AQL is 2.5, adjusted for product and end use.
- Minor defects: cosmetic flaws or deviations that do not affect function or safety. AQLs often run 4.0 or higher.
Because each category has its own AQL, the same lot can pass on minor defects and still be rejected for too many major ones. That layered structure is the point.
Reading the Tables
The lookup is a two-step sequence, and it never changes.
First, use the lot size and inspection level to find the sample size code letter in Table I. A lot of 500 units at General Level II, for example, points to code letter H.
Second, take that code letter into the correct master table (Table II-A covers single sampling under normal inspection) and find its row. Read across the row to the column for your specified AQL. The cell gives you three numbers: the sample size (n), the acceptance number (Ac), and the rejection number (Re). If the sample contains defects at or below Ac, accept the lot. If defects reach Re, reject it.
Sometimes the cell shows an arrow instead of numbers. A downward arrow means move to the next row and use its plan; an upward arrow means move up. Arrows appear where the statistical math does not produce a meaningful plan at that specific combination, and they route you to the nearest valid one. Follow the arrow until you land on actual numbers.
Single, Double, and Multiple Sampling Plans
A single sampling plan draws one sample of a fixed size, counts defects, and produces an accept/reject decision. It is the most common and the easiest to run.
A double sampling plan builds in a second chance. The first sample is checked against a lower acceptance number and a higher rejection number. Clear passes and clear rejects are decided immediately. Results in between trigger a second sample, and the combined defect count from both samples is compared to a final set of accept/reject numbers. Double plans often inspect fewer total units than single plans because most lots resolve on the first sample.
Multiple sampling plans extend the same logic across as many as seven stages, each with its own sample size, cumulative acceptance number, and cumulative rejection number. Total units inspected drop further for very good or very bad lots, but the bookkeeping across stages is unforgiving. Master tables provide columns for each stage with precise triggers.
Switching Rules
Z1.4 is not a one-off plan. The switching rules automatically tighten or loosen inspection based on recent lot history, so the system adapts to how the supplier is actually performing. 1ASQ. ANSI/ASQ Z1.4 and Z1.9 Sampling Plan Standards for Quality Control
Normal to Tightened
All inspection begins under normal. If two of five consecutive lots are rejected while on normal inspection, the standard requires a switch to tightened. Tightened plans use the same sample size but lower acceptance numbers, so borderline lots are less likely to pass.
Tightened Back to Normal
Returning to normal requires five consecutive lots accepted under the tightened criteria. Because those criteria are stricter, the supplier has to show genuine improvement to earn the switch back.
Normal to Reduced
Reduced inspection becomes available when a supplier demonstrates consistently good quality. The qualifying conditions include at least ten consecutive lots accepted under normal, a cumulative defect count that stays below a specified limit number in the tables, steady production, and approval by the responsible authority. Reduced plans use smaller samples, which lowers cost and time on both sides.
Reduced Back to Normal
Any single rejection while on reduced inspection sends the process straight back to normal. Production irregularities or changes in the conditions that justified reduced status do the same. The fast reversion is deliberate: it stops reduced inspection from hiding a slide in quality.
Discontinuation
The severe end of the switching system is full discontinuation. If the cumulative number of lots not accepted while on tightened inspection reaches five, acceptance procedures stop entirely. 2ANSI Webstore. ANSI/ASQ Z1.4-2003 – Sampling Procedures and Tables for Inspection by Attributes – Section 8.4 No further lots are accepted under Z1.4 until the supplier takes corrective action and demonstrates a capable process. When inspection resumes, it resumes under tightened rules, not normal.
Random Selection
The statistical basis for every number in the tables assumes the sample is drawn at random from the lot. Pulling units from the top layer of a pallet or from the most accessible cartons biases the sample and invalidates the acceptance decision. Random selection means every unit in the lot has an equal chance of being chosen, achieved through random number generators, pre-printed random selection tables, or systematic methods that cover the entire lot without pattern. Convenience sampling is the most common failure mode in real inspections, and it can render a perfectly constructed plan meaningless.
Producer’s Risk, Consumer’s Risk, and the OC Curve
Because you inspect a sample rather than the whole lot, two errors are always built into the system. Producer’s risk (alpha) is the probability that a lot which actually meets the AQL gets rejected because the random sample happened to contain more defects than typical. Consumer’s risk (beta) is the reverse: a lot worse than acceptable quality gets accepted because the sample happened to look better than reality.
The operating characteristic (OC) curve shows how a specific plan behaves. It plots the true percent defective on the horizontal axis against the probability of acceptance on the vertical axis. A steep curve discriminates sharply between good and bad lots. A shallow curve gives borderline lots close to a coin-flip chance of passing. Larger samples produce steeper curves, which is why Level III catches more borderline problems than Level I. Before agreeing to a plan with a supplier, check the OC curve at quality levels slightly worse than your AQL and confirm the protection is what you expected.
Z1.4 or Z1.9
Z1.4 covers inspection by attributes: each unit either passes or fails a defined criterion. Its companion standard, ANSI/ASQ Z1.9, covers inspection by variables, where the inspector takes actual measurements and evaluates the data statistically. 1ASQ. ANSI/ASQ Z1.4 and Z1.9 Sampling Plan Standards for Quality Control Z1.9 needs fewer units for the same statistical confidence because measurement data carries more information than a pass/fail count. It only applies, though, when the characteristic being measured follows a normal distribution, measurements are reliable, and one characteristic is being evaluated at a time. Z1.4 works with any attribute check, from visual cosmetic inspections to functional go/no-go tests, with no assumption about the underlying distribution.
Where the Results Carry Legal Weight
Z1.4 is not itself law, but several legal regimes rely on the type of plan it produces.
FDA-Regulated Medical Devices
Medical device manufacturers that use sampling plans must base them on a valid statistical rationale and document both the plans and any changes, and must establish procedures to confirm sampling methods remain adequate. 3eCFR. 21 CFR 820.250 – Statistical Techniques Citing Z1.4 satisfies the recognized-methodology requirement, but the manufacturer still has to document the AQL values, inspection levels, and lot definitions selected, and review those choices whenever processes or products change.
Federal Supply Contracts
Contractors under fixed-price federal supply contracts must maintain an inspection system acceptable to the government and tender only supplies that have been inspected and confirmed to conform. The government keeps the right to inspect at any point during manufacturing, contractors must keep inspection records available, and rejection or correction remains available for nonconforming supplies. Contractors also remain liable for latent defects or fraud discovered after acceptance. 4Acquisition.GOV. FAR 52.246-2 – Inspection of Supplies-Fixed-Price
Commercial Contracts Under the UCC
Between private parties, the Uniform Commercial Code gives Z1.4 results their teeth. Goods conform to a contract when they match the obligations the contract imposes. 5Cornell Law Institute. UCC 2-106 – Definitions When a purchase order specifies an AQL of 2.5 at General Level II, that specification becomes a contract term, and a lot that fails those criteria is nonconforming. The perfect tender rule then lets the buyer reject the entire shipment, accept the entire shipment, or accept some commercial units and reject the rest. 6Cornell Law School – Legal Information Institute. UCC 2-601 – Buyer’s Rights on Improper Delivery The reverse also applies. If the contract designates a Z1.4 plan and a lot passes under it, the buyer generally cannot reject based on a general sense that quality feels off. The plan becomes the agreed method for determining conformity, and its results carry contractual weight.