Apparel Inspection Sampling: How AQL 2.5 Works for Garment Orders — and When It Doesn’t

For most garment orders, inspecting every piece is too slow and too expensive. That is why buyers and factories use sampling: check a representative part of the order, and use the result to accept or reject the whole lot. The most common framework is AQL, or Acceptable Quality Limit. This guide explains it in plain terms, with a practical checklist you can use before your next inspection.

What AQL actually means

AQL is not a promise that a shipment is defect-free. It is a statistical agreement about how many defective pieces are tolerated in a sample before the lot is rejected.

In apparel, AQL 2.5 for major defects is a common starting point. It means that, at the agreed sample size, the number of major defects found must stay at or below the acceptance number for that sample. Find more than that, and the lot fails — even if most pieces are fine.

Three defect classes matter, and they should be defined before inspection starts:

• Critical defects: the garment is unsafe, unlawful to sell, or completely unusable. A broken needle left in a seam is a classic example. The usual tolerance is zero.
• Major defects: the garment is visibly wrong or will likely be returned — a wrong size, a broken zip, a large stain, a misaligned print.
• Minor defects: small flaws a customer may not notice, such as a loose thread end or a slight shade variation within an agreed range.

If these classes are not written down, the inspection becomes an argument. Agree them in the purchase order, the tech pack, or the inspection brief.

A simple worked example

Imagine a lot of 5,000 shirts. Under a common general inspection level, the inspector might draw about 200 pieces at random. With AQL 2.5 for major defects, the acceptance number for that sample might be around 10 major defects: find 10 or fewer, accept; find 11 or more, reject.

These numbers are illustrative, because the exact sample size and acceptance number depend on the lot size, the inspection level, and the sampling table you agree to use. Do not copy a number from another order without checking the table that applies to yours. The point is the logic: a larger lot does not mean inspecting everything, and a small sample does not mean a quick glance at the top carton.

How to make the sample fair

A sample is only useful if it is random. Ask the inspector to:

1. Pull cartons from different pallets and positions, not only the easiest ones to reach.
2. Mix sizes and colours in proportion to the order, unless a specific size or colour is the known risk.
3. Check garments as they will be shipped, including folding, polybags, labels, and carton markings.
4. Record the defect, the garment, and the size for every failure, with photos where possible.

If all the sampled pieces come from one carton, you have inspected one carton, not the order.

When AQL sampling is not enough

Sampling is a risk-management tool, not a guarantee. Use a different approach when:

• The risk is safety or compliance, such as children’s products, flammability, or restricted chemicals. Those need the relevant tests and records, not only a visual sample.
• A defect is systematic. If the same stitching error appears in every sampled piece, the problem is in the process, not in chance.
• The order is small or very high value. For a short run of premium garments, 100% inspection may cost less than a rejected shipment and a lost customer.
• You are inspecting after packing. Once goods are sealed, a failed inspection is much more expensive to fix.

In those cases, agree the alternative in advance: a higher inspection level, 100% checking of a critical point, or a pre-production sample that locks the standard.

What a good inspection report should contain

Before you pay for an inspection, ask what the report will show. A useful report includes the lot size, the sample size, the AQL and acceptance numbers used, the defect list by class, measurements against the agreed tolerance, photos of failures, and a clear accept or reject recommendation. A report that only says “quality is OK” is not evidence you can act on.

Buyer checklist before inspection day

• Defect classes and AQL levels agreed in writing.
• Approved sample, tech pack, and measurement tolerance available to the inspector.
• Random sampling method agreed, including size and colour mix.
• Rules for re-inspection after rework: who pays, what is rechecked, and how many times.
• A named contact at the factory who can authorise opening cartons and rework.

AQL does not remove quality risk, but it does make quality decisions testable. When the standard, the sample, and the acceptance rule are agreed before production finishes, both buyer and factory know what “good enough” means — and a failed lot is a decision, not a dispute.

Related reading: Garment AQL Inspection Limits — what an AQL result can and cannot prove about a shipment.