AQL Inspection for Textiles: Sampling and Defect Classes

Anyone buying clothing in larger quantities will sooner or later come across terms like “AQL 2.5/4.0”. This refers to a sampling method. Instead of looking at every piece, inspectors check a fixed number of randomly drawn pieces and decide on the whole shipment based on the defects they find. This guide explains how the method is built, how to read a sample size from the tables and where its limits are.

What exactly does AQL mean?

AQL stands for Acceptance Quality Limit. It is the worst average defect rate that is still tolerated across a continuing series of lots. At AQL 2.5, that means an average of 2.5 defective pieces per 100. The basis is ISO 2859-1, which contains sampling plans for inspection by attributes. Each piece drawn is rated as either acceptable or defective; no measured value is carried forward.

The standard is designed for recurring deliveries, for example when the same item is reordered regularly. For a single, isolated lot there is a separate part, ISO 2859-2, which works with a limiting quality. In the apparel trade, the tables from part 1 are nevertheless often used for one-off orders too.

How do I read the sample size?

You need the lot size, meaning the number of pieces in the delivery being inspected, and the inspection level. With these two values you go through the tables of the standard in three steps.

• Lot size and inspection level give you a code letter in Table 1.
• For that code letter, Table 2-A (single sampling, normal inspection) gives the sample size.
• The same row shows the acceptance number Ac and the rejection number Re for each AQL value.

An example: 5,000 T-shirts at general inspection level II. This gives code letter L and a sample of 200 pieces. At AQL 2.5 the acceptance number is 10 and the rejection number is 11. If the inspectors find no more than 10 pieces with major defects, the lot is accepted for that defect class. From 11 pieces on, it is rejected. For minor defects at AQL 4.0, the limits for the same sample are 14 and 15.

For orientation, here are some values for inspection level II. Each line shows the sample size followed by Ac/Re for AQL 2.5 and for AQL 4.0:

• 281 to 500 pieces: inspect 50 pieces, 3/4 and 5/6
• 501 to 1,200 pieces: inspect 80 pieces, 5/6 and 7/8
• 1,201 to 3,200 pieces: inspect 125 pieces, 7/8 and 10/11
• 3,201 to 10,000 pieces: inspect 200 pieces, 10/11 and 14/15
• 10,001 to 35,000 pieces: inspect 315 pieces, 14/15 and 21/22

The sample grows much more slowly than the lot. With 500 pieces, every tenth piece is checked; with 10,000 pieces, only every fiftieth. Where the table shows an arrow instead of a number, you use the first plan in the direction of the arrow, together with its sample size. This mainly happens with small lots and very strict AQL values.

Which inspection levels are there?

ISO 2859-1 has three general inspection levels (I, II and III) and four special ones (S-1 to S-4). Unless something else is agreed, level II applies. Level I uses smaller samples, level III larger ones. The special levels get by with very few pieces and are meant for tests that take a lot of time or destroy goods. For textiles, that could be measuring every point of measure or a wash test.

For recurring deliveries, the standard also has switching rules. If two out of five or fewer consecutive lots are rejected, inspection switches to tightened, with stricter acceptance numbers. After a longer run of accepted lots, reduced inspection is possible under certain conditions. If you apply AQL to a single order only, these rules usually do not come into play.

What are critical, major and minor defects?

The standard does not say what counts as a defect for your product. Buyer and producer agree on that beforehand, ideally in a defect list with photos. Three classes are common. They are counted separately and each checked against its own limits.

• Critical defects can endanger people or make the goods unfit to be sold legally. Examples are a broken needle tip left in a garment, sharp edges on metal parts or drawstrings in the hood and neck area of children’s clothing.
• Major defects make a piece unsellable or are obvious to the end customer straight away. These include an open seam, a hole, a measurement outside tolerance, a clear colour difference between sleeve and front panel, a wrong size label or a crooked print.
• Minor defects deviate from the specification without a customer returning the piece because of them. Typical examples are untrimmed thread ends, slightly uneven stitch density in a hidden spot or a small mark that can be removed.

You will often see the combination 0 for critical, 2.5 for major and 4.0 for minor defects. This split is widespread, but the standard does not prescribe it. For children’s clothing or high-priced items, a stricter value for major defects can make sense, such as 1.5. The 0 for critical defects means that a single finding leads to rejection.

How a defect is classified also depends on where it is. A small weaving flaw in the lining of a jacket matters less than the same flaw on the chest of a white T-shirt. So define which zones of a garment count as clearly visible and which count as hidden.

What does a passed AQL inspection tell you?

It tells you that this sample did not contain more defects than allowed. It says less about the actual defect rate of the whole shipment than many people assume, because chance plays a part. If you work through the example with 200 pieces and an acceptance number of 10, you get roughly this picture:

• A lot that is actually 2.5 % defective is accepted in about 99 out of 100 cases.
• If the true defect rate is 5 %, twice the AQL, the lot is still accepted in about 58 out of 100 cases.
• Only at around 10 % defective goods is acceptance practically ruled out.

Critical defects behave in a similar way. If 0.5 % of the pieces in a lot have a critical defect, a sample of 200 pieces will, by calculation, find none of them in a little over a third of cases. For safety-relevant features a sample is therefore often not enough. What helps there is a full check of that specific feature or controls within the process, such as a needle detector that every finished piece passes through.

Apart from the statistics, the method has other limits:

• Only what is in the defect list gets checked. A flaw nobody described beforehand can easily slip through.
• Colour fastness, shrinkage after washing or harmful substances cannot be seen on a garment. These need laboratory tests.
• The result only holds if the pieces were genuinely drawn at random from the finished, packed goods. Pieces set aside by the producer say nothing about the lot.
• An accepted lot can still differ from the approved sample if the reference in the defect list was vague.

What should you agree before ordering?

If you want to work with AQL, it belongs in the specification before production starts. It is hard to negotiate afterwards, because by then the producer has long been working to their own assumptions. These points should be settled in writing:

• a defect list with photos that assigns every defect to a class
• a size chart with measuring method and permitted tolerance for each point of measure, as described in the tech pack guide
• inspection level and AQL value for each defect class
• how pieces are drawn, for example from randomly chosen cartons of the goods ready for shipping
• what happens on rejection: sorting, rework, re-inspection and the effect on the delivery date
• which laboratory tests are commissioned separately from the visual inspection

What makes sense after a failed result is covered in our article on a failed pre-shipment inspection. As a template for the final check, you can download our PSI checklist and adapt it to your item.

How does DS Trading check the goods?

AQL is not our standard. Our own team in China checks on site at the producer, twice as standard: during production and after production, before the goods are shipped. The benchmark is the sample you approved in writing, together with your specification. The check during production has a practical reason. A defect our team finds on the first finished pieces can still be fixed by the producer before it runs through the entire order.

If you have your own acceptance criteria, for example specific AQL values or an inspection level from your quality manual, describe them in the specification. We discuss them with you before the order and work out how they fit the item and our on-site checks. You can also waive the checks, which makes the order cheaper.

How the two checks work is explained on our quality control page. The final check before shipping is covered in our guide to pre-shipment inspection. If you are planning your own collection, you will find the whole route from sample to delivery under importing private label clothing.