AQL Inspection for Brand Owners: What 2.5 Actually Means

Almost every article on AQL is written for a professional QC buyer who already knows what a sampling plan is. This one is written for the brand owner signing the purchase order, who needs to know what number to write, when to book, what it costs, and what happens when a lot fails. It sits within sourcing apparel direct from factories, and it starts where the sampling calendar from proto to TOP ends: with a sealed sample the bulk will be judged against.

What AQL 2.5 actually means, and what it does not

AQL stands for Acceptable Quality Limit. It is defined in ANSI/ASQ Z1.4, the American standard, which is equivalent to ISO 2859-1 internationally. Both describe the same thing: a statistical procedure for deciding whether to accept or reject a whole lot after inspecting a sample of it.

The common misreading is that AQL 2.5 promises no more than 2.5% defective goods. It does not. AQL 2.5 is the defect level at which a lot will be accepted with high probability by the sampling plan. It describes the behaviour of the decision rule, not the condition of your inventory. A lot can pass at AQL 2.5 with a sample defect rate well above 2.5%. The standard is a commercial risk threshold applied to a sampling decision, not a quality target and certainly not a warranty.

That distinction matters because brands routinely write "AQL 2.5" into a PO believing they have bought a 97.5% good-goods guarantee, then discover the returns rate says otherwise.

How an ANSI/ASQ Z1.4 sampling plan is actually built

Four inputs produce the accept/reject decision.

1. Lot size. The units offered for inspection, usually the whole order or one style-colour.
2. Inspection level. General Inspection Level II is the apparel default. Levels I and III trade sample size against confidence. Special levels S-1 to S-4 are used for slow or destructive checks, such as measurement against points of measure or seam strength.
3. Sample size code letter. Lot size plus inspection level maps to a letter, which maps to a sample size.
4. Accept/reject numbers. The code letter plus the AQL gives Ac (maximum defects to accept) and Re (defects at which the lot is rejected).

Sample sizes and accept/reject numbers at General Inspection Level II, single sampling, normal severity. These are the published ANSI/ASQ Z1.4 values; verify them against your own copy of the Z1.4 or ISO 2859-1 table when you write them into a contract, so both you and the factory are reading the same row.

Notice that sample size grows far more slowly than lot size: doubling your order does not double the number of pieces anyone looks at. That is the whole point of acceptance sampling, and also its limitation.

Worked example: a 5,000-piece tee order at AQL 2.5

• Lot size 5,000 pieces, so the 3,201–10,000 row applies.
• General Inspection Level II gives code letter L, sample size 200 pieces.
• At AQL 2.5 the accept number is 10 and the reject number is 11.
• The inspector randomly draws 200 pieces from packed cartons across the lot.
• If 10 or fewer carry major defects, the lot passes. At 11, it fails.

10 ÷ 200 = 5.0%. A lot can be 5.0% defective in the sample and still be accepted at AQL 2.5. Apply that rate to 5,000 pieces and you are looking at potentially 250 units your customers may see as faulty. That is the number nobody puts in the brochure, and it is why serious brands pair the AQL with a specific defect list and a during-production check rather than relying on the final number alone.

Critical, major and minor defects in apparel inspection

Defect classification and the standard apparel AQL combination. Each class is judged independently against its own accept/reject numbers.

The independence rule matters commercially: a lot can pass on minors and fail on majors, and the reverse. You are running three tests, not one.

Critical at AQL 0 is where compliance and quality meet. A missing or wrong origin label is a critical defect and also a violation of the FTC's textile labeling rules, which require the specific country name and, for garments with a neck, placement at the inside centre of the neck. See US apparel labeling requirements for what has to be on the label before an inspector can pass it.

When to book: during-production versus final random inspection

There are two bookings that matter, and most brands only make the second one.

• During-production inspection (DUPRO), booked at roughly 20% of output. Some units are finished, the line is running, and the systemic problem is still correctable. If the neck rib is stretching out or body and sleeve are shading differently, this is where the fabric roll can be pulled and the operation reset.
• Final random inspection (FRI), booked at 80% or more of goods packed. This is the industry default and a trap when used alone. At 80% packed the order is made, and every remedy from here is a sort, a rework, a discount or a rejection.

For a first order with a new factory, book both. For a repeat order on a proven style, FRI alone is usually defensible.

What third-party garment inspection costs

Apparel inspection pricing per man-day, 2026. One man-day typically covers one style at one factory; multi-style or multi-factory orders consume more.

Put that against the order. A 5,000-piece tee programme at $4.30 FOB is a $21,500 order. One QIMA man-day at $320 is 1.5% of order value. Two inspections, DUPRO plus FRI, are 3.0%. That is the cheapest insurance on the entire purchase order, and it is why the pre-shipment inspection cost belongs in your landed cost for apparel imports build rather than being treated as an optional extra.

The rework economics of an AQL failure

Rework and re-inspection after a failed inspection typically cost 4–9% of order value. Applied to the same $21,500 order:

That is 33–38% of the FOB value of the order, from one failed inspection on a basic tee. The air-freight line does the damage: a 180 gsm tee weighs about 0.23 kg packed, and at roughly $6.00/kg (Asia to North America, August 2026, volatile) air freight runs about $1.38 a unit against $0.15–0.30 by ocean. Freight rates move weekly, so treat that line as indicative of magnitude rather than a quote. The lesson is not "inspect harder": it is that the DUPRO booking, which costs one man-day, is what prevents both the 4–9% and the air freight.

The test methods brands should be specifying, and where they belong

An AQL inspection is a visual and dimensional check. It does not tell you whether the fabric will survive washing. That is what the GPT (garment performance testing) sample is for, and it must be specified in the tech pack and the PO with method numbers, not adjectives.

Colourfastness minimums to write into the purchase order. Grey scale runs 1–5, where 5 is no change; report both the colour-change scale and the staining scale.

Dimensional stability and skew. Specify AATCC TM135 for the North American convention (top-loading machines) or ISO 6330 for the European convention (front-loading, higher mechanical action). The results are not interchangeable, so name one, and write the AATCC call-out in full, for example 135-3-IV-Ai, which fixes wash number, temperature and drying method. Condition specimens at 20°C and 65% relative humidity for at least 24 hours before measuring, per ASTM D1776 / ISO 139.

The industry norm is ±3% length and width for cotton knits and ±2% for wovens. Untreated 100% ring-spun cotton jersey shrinks around 5.0–7.5%, so the fabric must be compacted (mechanically pre-shrunk) to get inside that tolerance; if you do not specify compacting, you will receive untreated fabric that meets your GSM and fails your shrinkage. For spirality and skew, measure per ISO 16322 or AATCC TM179 and accept at 5% displacement or less on knits, with more than 4° or 5% treated as a major failure.

Two points connect back to the rest of the file. First, shade variation between panels is a major defect, which means your colour tolerance has to be written in a form an inspector can apply: see Pantone and Delta E colour approval. Second, whether a fabric can hit these numbers at all is set at specification time, in the yarn, structure, dye method and finish: see reading and writing a fabric spec.

How to set up apparel AQL inspections

1. Write the AQL terms into the purchase order, not the email thread: standard (ANSI/ASQ Z1.4 or ISO 2859-1), AQL per class, inspection level, sampling plan type and severity.
2. Build a defect list specific to the style. A generic garment inspection checklist misses style-specific failure modes, and the modes that matter differ by product category, which is the argument in why one factory cannot make your whole range. Classify each listed defect and send the list with the sealed sample.
3. Book the during-production inspection at about 20% of output. This is the booking that saves the money.
4. Specify the measurement sample separately: commonly 8–15 pieces per size at Special Inspection Level S-2 or S-3, measured against the tech pack points of measure and tolerances, with actual values reported rather than a tick.
5. Book the final random inspection at 80% or more packed, with 5–7 working days notice.
6. Define in advance what happens on a fail: who sorts, who pays, who re-inspects, and who on your side can accept, discount or reject within 24 hours.

Definitions for AQL, DUPRO, crocking and grey scale are collected in the Yarnstick glossary of sourcing terms.

Frequently asked questions