AQL garment inspection uses an agreed sampling plan to support a lot decision. An AQL number is not the percentage of goods to inspect or a shipment defect allowance. To use AQL 2.5 or another agreed index, define the lot, inspection unit, level, sampling scheme and decision rules together.
Acceptance sampling uses a sample to support an accept-or-reject decision about a lot. It is different from estimating the exact quality level of every unit in that lot. Sampling involves the possibility of accepting or rejecting lots based on the sample outcome; it does not provide a zero-defect guarantee.
ISO 2859-1:2026 is the current published edition reviewed for this guide. It describes AQL-indexed sampling schemes for inspection by attributes. Record the standard and edition actually agreed for the order, and use its applicable tables and rules when selecting the plan.
| Input | Question to settle |
|---|---|
| Inspection lot | Which order, styles, colors, materials and production grouping form the lot? |
| Lot size | How many defined inspection units are included? |
| Inspection unit | Is one unit a single garment, a complete set or another expressly agreed object? |
| Standard and scheme | Which standard/edition and single, double or other sampling arrangement apply? |
| Inspection level | Which level is selected for this inspection? |
| Severity and switching | Is inspection normal, tightened or reduced, and how do the scheme's switching rules apply? |
| Defect categories | Which findings are critical, major or minor under the agreed definitions? |
| Sampling and decision | What sample size and acceptance/rejection numbers result from the plan? |
| Follow-up | What happens after rejection, correction, sorting or reinspection, and who authorizes release? |
For Jinshi's normal sales-unit basis, a complete two-piece set is one unit. Inspecting a sampled set includes both pieces and their pairing. If a buyer requires a different inspection-unit basis, record it separately so defect counts and quantities remain understandable.
The quality plan needs explicit unacceptable conditions and disposition rules for safety-related findings. Examples for review include a dangerous sharp part or a confirmed hazardous attachment failure. The category depends on the actual product, intended use and applicable requirements.
These can materially affect function, use or saleability under the agreed specification. A zipper that does not work, a significant open seam or a materially incorrect size can be examples. Use the buyer's definition and illustrated reference where available.
These are lesser departures within the agreed defect framework, such as a small appearance fault that does not materially affect intended use. A feature should not be called minor merely because it is small; consequence and the approved specification matter.
Jinshi uses AQL 2.5 for major defects as a discussion starting point, with minor and safety-critical criteria established separately. The agreed plan, rather than the number alone, determines how the inspection operates.
Unit-by-unit checks can identify appearance, key function and labeling issues. Sampled lot inspection supports the agreed shipment decision. Laboratory work assesses specified material, performance or safety properties using its own specimens and methods.
All three can be part of the same project. A passed sampled inspection does not replace a flammability report, and a laboratory result does not count the correct size assortment in every carton. See the production inspection stages.
An observed nonconforming-unit rate is nonconforming units ÷ inspected units × 100 for a stated population and period. Define whether the measurement concerns first inspection, reinspection, customer receipt or retail returns.
For an illustrative month, suppose 10,000 completed sales units receive their first finished-product inspection and 27 have at least one defect requiring repair or rejection. The first-inspection nonconforming-unit rate is 0.27%.
Count a unit once in that total even if it has several defects, while recording each defect type in the separate analysis. If either piece of a set is nonconforming, the complete set counts as one nonconforming sales unit. Reinspection results should not overwrite the original first-inspection record.
Jinshi's proposed management target is a monthly first finished-product inspection nonconforming-unit rate below 0.3%. This is a target, and 27/10,000 is a calculation example; neither is a claim of achieved historical performance or permission to ship those affected goods.
No. It is an index used within the sampling scheme. It does not set a guaranteed maximum defect percentage for each lot or authorize known nonconforming goods for shipment.
A fixed percentage is not the same as an AQL-indexed plan. Select the sample and decision numbers from the agreed scheme and its inputs.
No. It defines the decision when such a finding is observed in the inspected sample. The safety and production-control plan still needs to address the product and the full relevant scope.
Identify the finding and affected scope, agree sorting or correction, and perform the required verification or reinspection. Release follows the agreed decision authority and conditions.
ISO 2859-1:2026 and NIST: acceptance sampling. Reviewed 8 October 2026.
