AQL sampling for fasteners is a lot-acceptance method: an inspector selects a defined sample, checks specified characteristics, and applies a predetermined accept-or-reject rule. It is not a promise that every accepted lot contains exactly the AQL percentage of defects, and it does not replace a drawing, test standard, or process-control program. To use the method well, the buyer must identify the lot, classify defects, choose the governing sampling plan, and record the acceptance number before inspection begins. Random, traceable selection matters as much as the number of pieces checked.
Acceptance sampling uses evidence from a sample to make a decision about a submitted lot. It sits between inspecting every unit and inspecting none. That can be useful when complete inspection is expensive, time-consuming, or destructive. The decision is a disposition of that lot under the agreed plan, not a statistical estimate that certifies every bolt, nut, or screw in the shipment.
A buyer should distinguish acceptance sampling from process control. A sample can find a problem and help decide whether a lot should be accepted, rejected, sorted, or investigated. It does not reveal when a process started drifting or identify every cause. Suppliers still need process controls and corrective action. For safety-related or critical characteristics, customer requirements may call for a different verification method or more extensive inspection.
Acceptance sampling also does not define what a defect is. The drawing, product specification, customer standard, or approved control plan must say which characteristics are checked and what constitutes a nonconforming unit. Thread damage, wrong pitch, incomplete coating, incorrect marking, and mixed lots are different conditions and may have different consequences.
A lot must be a defined group of product made or processed under the conditions relevant to the inspection plan. Fasteners with different part numbers, sizes, grades, materials, coatings, manufacturing runs, or heat/lot identifiers should not be combined simply to produce a larger inspection population. If the specification defines lot formation, use that definition. If it does not, state the rule on the purchase order or quality plan and keep each supplier lot identifiable.
After defining the lot, select units in a way that avoids bias. Samples taken only from the top of a carton or from the easiest bins may miss variation across packages or production time. Use the plan’s prescribed random method and preserve package or container identities. If the shipment contains several sub-lots, the inspector should know whether the plan applies to each sub-lot separately or to the combined population under an approved rule.
Keep acceptance sampling separate from destructive testing. A tensile test, proof-load test, coating test, or chemical analysis may need a distinct number of specimens and a standard-specific method. Do not assume the general incoming AQL sample automatically satisfies those product-test requirements. Record the sample selection method, sample identity, and the test method used for each purpose.
In ANSI/ASQ Z1.4, the Acceptance Quality Limit (AQL) indexes a family of sampling plans used with switching rules over a continuing series of lots. It describes a level around which the plan is designed, but an accepted lot can still contain nonconforming units. The chance of accepting or rejecting lots depends on the actual quality level, sample size, acceptance number, and chosen plan. The operating-characteristic curve describes that probability trade-off.
Do not tell a supplier that “AQL 1.0 means one percent of the order may be defective.” That confuses a plan parameter with a guaranteed allowance. Nor should a buyer select a familiar AQL value without deciding how the characteristic affects safety, fit, assembly, appearance, or function. The contract should specify defect classes, inspection level, the standard and edition, plan type, and any special customer criteria.
Different standards and revisions can use different tables, terminology, and rules. ANSI/ASQ Z1.4 covers attributes sampling; Z1.9 covers variables sampling under its assumptions. ISO 2859-1 is a related international standard, but do not substitute tables or rules without authorization. Obtain the current edition specified by the customer and use its complete switching and disposition rules.
Attributes inspection records a category or count, such as whether a thread gauge result passes or how many fasteners have a visible defect. Variables inspection records a measured value, such as length or diameter, and evaluates it under the selected plan. Variables plans may use measurement information more efficiently in suitable cases, but they require an appropriate distribution assumption and consistent measurement system.
For a bolt lot, an attributes plan might count parts with wrong markings or a failed functional thread gauge. A variables plan might use actual length readings when the controlling standard and plan permit that approach. If characteristics have different tolerances, distributions, failure consequences, or measurement methods, do not force them into one generic decision. Use the applicable characteristic-by-characteristic plan and product standard.
| Plan input | Fastener example | What the buyer must define | Record to retain |
|---|---|---|---|
| Lot identity | One part number, size, finish and supplier lot | Which packages or sub-lots are included | Part number, drawing revision, supplier and lot IDs |
| Characteristic | Thread fit, length, coating, marking or head geometry | Method, datum, limits and applicable standard | Characteristic ID and actual observations |
| Defect class | Critical safety feature, major fit issue or minor cosmetic issue | Customer-approved classification and consequence | Defect code and disposition rule |
| Sampling plan | Single, double, multiple or variables plan | Standard, edition, inspection level and AQL/index | Sample size, acceptance/rejection numbers and switching state |
| Decision and action | Accept, reject, sort, hold or investigate | Authority for concessions, retest and corrective action | Decision, approver, date and linked corrective action |
The table is a planning checklist, not a substitute for the controlled sampling tables. For the statistical meaning and limitations, see the NIST Engineering Statistics Handbook section on acceptance sampling. The ASQ sampling overview explains randomization and sampling error, while ASQ’s Z1.4 and Z1.9 standards page summarizes the standards and their intended uses.
Critical, major, and minor are useful buyer categories only when the customer or contract defines them. A missing or malformed safety feature may require a different rule from a cosmetic mark that does not affect fit or performance. State the criteria in measurable language and give inspectors photographs or drawings where visual acceptance is involved. Avoid vague definitions such as “workmanship must be good.”
Also decide whether the unit of count is a defective fastener or each individual nonconformity. A single fastener may have more than one defect. The selected standard’s terminology and table rules govern how defects are counted. If different criticalities are counted separately, keep separate records instead of combining them into one total that hides the most serious result.
For requirements that demand zero acceptance, state that explicitly and identify the affected characteristic. Do not infer a zero-acceptance plan from a low AQL label. Likewise, do not loosen a critical characteristic’s criteria after seeing a sample result. Any concession should go through the authorized engineering or customer approval process and remain attached to the lot record.
Verify the purchase order, drawing revision, product identification, lot paperwork, package quantity, and material or finish records. Confirm that inspection equipment is suitable and in calibration. Select samples according to the plan and preserve the original lot segregation. Perform each check using the stated method and record actual values or pass/fail results rather than recording only “inspected.”
Compare results to the plan’s acceptance and rejection numbers. If a lot is rejected, follow the written rule for containment, sorting, return, rework, and resubmission. A resubmitted lot may require a fresh identity and a different inspection severity under the governing standard. Do not combine reworked and unaffected material without explicit traceability and an approved rule. Keep both the first result and later disposition in the record.
Link inspection results to the supplier and lot certificate. TNHO’s hex and square-head bolt family illustrates fastener geometries buyers may inspect, but actual acceptance criteria come from the drawing and purchase requirements. See the fastener first-article inspection guide for initial qualification, the dimensional inspection guide for tool selection, and the material certificate guide for lot evidence.
The ASQ Statistics Division video below shows how a Z1.4-style attributes sampling plan is selected and how operating-characteristic curves help describe producer and consumer risk. It uses an older edition as a teaching example. Use the current edition named in the contract for actual fastener inspection, including its tables, switching rules, and terminology.
Video: Acceptance Sampling Using Z1.4. The example explains acceptance-sampling concepts; it does not select an AQL or acceptance number for a particular fastener order.
Specify the lot definition, characteristic list, defect classifications, governing inspection standard and revision, attributes or variables method, inspection level, sampling plan type, AQL or other plan index for each class, sample-selection rule, acceptance/rejection numbers, retest or resubmission policy, and reporting format. State when the customer must be notified and who can approve a deviation. Attach the released drawing and identify features that cannot be accepted by sampling alone.
For a custom screw or bolt, also define the material, heat or coating lot, thread class, functional gauge, marking, packaging, and traceability expected. Keep the sampling plan aligned with the process risk and customer contract. AQL tables cannot correct a poorly defined drawing, mixed lots, biased sampling, or a measurement system that cannot resolve the tolerance.
No. AQL is an index used with a specified sampling plan and switching rules. The plan controls lot-acceptance probabilities; an accepted lot can still contain nonconforming units.
Not automatically. The customer or quality plan should classify characteristics by risk and state the applicable decision rule. Safety, fit, function, and cosmetic features may need different criteria.
No. Acceptance sampling supports a lot disposition. Process control monitors production behavior and helps identify when a process changes. A receiving sample does not replace supplier process controls.
Only if the plan defines that selection method. A representative random sample should follow the governing procedure and preserve package and lot identity; convenience selection can bias the result.
TNHO 
