AQL final inspection for activewear: what a pass proves

What does AQL 2.5 actually mean?

It means the sampling scheme is designed to accept most lots from a process averaging 2.5% defective or better. ISO 2859-1 defines the acceptance quality limit as the worst tolerable process average when a continuing series of lots is submitted for sampling. It is a parameter of the plan, not the largest share of faulty garments your shipment may contain.

The standard itself notes that a lot as bad as the AQL can still be accepted with fairly high probability, and cautions that naming an AQL does not entitle a supplier to ship nonconforming items knowingly. ISO published the third edition, ISO 2859-1:2026, in January 2026. Many inspection firms work from the US counterpart, ANSI/ASQ Z1.4; SGS states that its final random inspections sample to Z1.4. Name the standard and edition in your order.

The scheme also assumes a continuing series of lots, with switching rules that tighten inspection when results deteriorate. A single first order never gives those rules a chance to work. For isolated lots, ISO 2859-2:2020 offers plans indexed by limiting quality instead.

How many garments will the inspector check?

For lots of 91 to 10,000 pieces at the default level, between 20 and 200 garments, set by the lot size and the inspection level. ISO 2859-1 uses general level II unless something else is specified; level III draws larger samples and level I smaller ones. The figures below are single sampling plans for normal inspection at level II. A lot fails as soon as the count of defective garments exceeds the pass figure.

The standard explains that the information a sample gives depends on its absolute size, not on its share of the lot. Twenty garments are a fifth of a 100-piece order, yet by our calculation a 100-piece lot that is 10% defective passes its plan about 36% of the time, while a 2,000-piece lot at the same defect rate passes the 125-piece plan only about 5% of the time.

What can a passed inspection not tell you?

It cannot vouch for the garments nobody opened, and it cannot see properties that only appear after washing or testing. Take a hypothetical lot of 2,000 tops containing 100 defective pieces, or 5%: by our calculation it passes the AQL 2.5 plan above about 72% of the time, and at 8% defective it still passes about one time in five.

Defects also cluster. One size cut from a faulty marker, one colour dyed off-shade or one operator’s skipped stitches can sit in a few cartons that a purely random draw misses. Ask for the sample to be drawn across every colour, size and carton range in proportion to the order.

On site, an inspector checks appearance, workmanship, measurements, quantities, labels and packing. Colour fastness to washing (ISO 105-C06), dimensional change after laundering by ISO 6330 procedures, pilling, fibre composition and restricted substances need laboratory reports on the bulk fabric, agreed before production starts.

Which activewear defects are critical, major or minor?

Critical defects make a garment unsafe or unlawful to sell, major defects make it unsaleable or likely to be returned, and minor defects are cosmetic. Define each class in the inspection protocol, not on the factory floor. ISO 2859-1 leaves the AQL to the contract; many apparel buyers accept no critical defects and set AQL 2.5 for major and 4.0 for minor defects, a convention rather than a requirement of the standard. The examples below are a typical starting point for knitted performance wear.

What should the inspection protocol say before you book?

It should define the lot, the sampling plan, the references, the product-specific checks, the packing requirements and what happens after the result. SGS describes final random inspection as an on-site check made when production is complete and 80–100% export packed, against the reference sample, confirmed specification, purchase order and packing instructions. Each of those references has to exist in writing before the inspector arrives.

SenLin’s published measurement tolerance is 1–2 cm for performance tees, shorts and pants and 1–3 cm for fleece styles such as sweatpants and midlayers, which gives an inspector a written limit rather than a judgement call. SenLin provides a bulk inspection record; any third-party inspection, its scope and the AQL are agreed per order, and SenLin does not publish an AQL level.

• Lot: style, colours and quantity, and whether several styles count as one lot.
• Sampling: standard and edition, inspection level, AQL for each defect class, and a draw across colours and sizes.
• References: approved sample code, specification revision, measurement chart with tolerances, artwork and label approvals.
• Activewear checks: seams stretched by hand, shade match within sets, heat-transfer and reflective trim adhesion, waistbands relaxed and extended.
• Packing: carton count, size ratio, carton marks, barcodes and polybag requirements.
• Outcome: who receives the report, who may release goods marked for the client’s decision, and who pays for re-inspection.

What happens when a lot fails?

A failed lot should be sorted, not patched. ISO 2859-1 says a non-accepted lot is not resubmitted until every item has been re-examined and nonconforming items removed, replaced or corrected. The responsible authority then decides whether re-inspection is normal or tightened, and whether it covers every defect class or only those that caused the failure.

Replacing only the garments the inspector found leaves the rest of the lot unchanged, because the sample stood in for the whole order. Agree three outcomes in advance: sort and re-inspect, accept with a written concession, or reject. Keep the report, photographs and sampling plan with the purchase order; they are your evidence for any later claim.