# What is AQL sampling? A simple explanation for importers
If you import goods from China, you will run into the question of what is AQL sampling sooner or later. It usually shows up on an inspection quotation or in a discussion about your quality requirements. The short answer: AQL sampling is a statistical method for checking product quality. Instead of inspecting every unit in your order, an inspector checks a representative sample and uses the results to decide whether the whole lot passes or fails.
Most third-party quality inspections in China run on this one idea. Once you understand what is AQL sampling and how the method works, you can set sensible quality requirements, read inspection quotes without confusion, and avoid the two classic mistakes: paying to check everything (slow and expensive) or checking nothing and hoping for the best.
What is AQL sampling, and what it means for importers
AQL stands for acceptable quality limit. In plain terms, it is the worst quality level you are willing to accept for a production lot, expressed as a percentage of defective units. Agree an AQL of 2.5 for major defects and you are saying: if more than 2.5% of the inspected sample has major defects, the lot fails.
So what is AQL sampling in practice? Two things combined: a sampling plan (how many units get pulled from the lot) and an acceptance rule (how many defects are allowed before the lot is rejected). Both come from published tables in the standard ISO 2859-1, known in the US as ANSI ASQ Z1.4. The inspector does not invent the sample size on the day. The lot size and your chosen settings point to a row in the table, and the table fixes how many pieces to inspect and the maximum defects permitted.
That structure is the point. Two inspectors, on two different days, working from the same AQL settings, pull the same sample size and apply the same pass or fail rule. Without it, "check the quality" means whatever the inspector feels like that morning.
One thing the question of what is AQL sampling often hides: the method accepts that a small number of defects can exist in a passed lot. It is not a promise of zero defects. It is a statistical agreement about how much imperfection you will tolerate, which is why choosing the right settings matters as much as booking the inspection at all.
How AQL sampling works, step by step
The process runs the same way at almost every inspection company. Here is what is AQL sampling from the inspector's point of view.
First, the lot has to be ready. For a pre-shipment inspection, the standard condition is 100% of the goods produced and at least 80% packed. Sampling only works when the inspector can draw at random from a finished lot. If production is still running and cartons are being sealed while the inspector works, the sample is not representative and the result is not reliable.
Second, the inspector determines the sample size. This comes from the AQL tables and depends on the total lot size plus the inspection level you chose (more on levels below). A larger order gets a larger sample, but the proportion shrinks. A lot of 1,000 units might have 80 pieces inspected, while a lot of 35,000 might have 315. The math keeps statistical confidence roughly constant without the work growing linearly.
Third, the inspector pulls the sample at random from across the lot. Not from the top of the nearest cartons. Random selection is what lets the sample stand in for the whole shipment. A disciplined inspector opens cartons spread across pallets and production batches.
Fourth, each sampled unit gets checked against your requirements: appearance, dimensions, workmanship, function, safety, quantity, packaging, labeling, and barcodes, compared against the approved sample and your spec sheet. Every defect found is classified, typically as critical, major, or minor, and logged with photos.
Fifth, the defect counts go up against the acceptance numbers for your AQL settings. Within limits, the lot passes. Over the limit, it fails. Many firms also use a middle outcome, conditional pass, where the lot is accepted subject to rework of the defective portion and sometimes a re-inspection.
The verdict is what you pay for. The report is evidence; the pass or fail is the decision it supports.
AQL inspection levels: I, II, and III
One part of what is AQL sampling trips up new importers: there are two separate dials, and people mix them up. The inspection level (I, II, or III) controls the sample size. The AQL value (1.0, 2.5, 4.0, and so on) controls how many defects are tolerated. Different jobs.
Level II is the default and the right choice for most consumer goods. Level I is reduced inspection: smaller samples, for when you have a long clean history with a supplier and want to spend less on routine checks. Level III is tightened inspection: larger samples, for high-value goods, new suppliers, or after a quality failure, when you want more confidence.
Changing the level changes how many units get inspected, not how strict the pass rule is. A Level III inspection of the same lot with the same AQL values checks more pieces, which makes the result more reliable, but the allowed defect percentage stays the same. Want a stricter quality bar? Change the AQL value, not the level. Getting these two dials straight is half of understanding what is AQL sampling at a practical level.
Where AQL sampling fits in your inspection plan
AQL sampling is a method, not an inspection type. It is the counting rule used inside the four standard inspection types done in China: pre-production inspection (PPI), during-production inspection (DUPRO), pre-shipment inspection (PSI), and container loading supervision.
In practice, what is AQL sampling most often doing? Sitting inside the pre-shipment inspection. PSI is the most common inspection importers book, because it happens when the goods are finished and it is your last chance to catch problems before you pay the balance. The inspector applies AQL sampling to decide how many cartons to open and how many units to test, then gives you a pass or fail before the goods leave the factory.
DUPRO uses the same sampling logic mid-production, usually when 20-60% of the order is complete. The sample comes from finished units on the line. PPI is different in nature: it checks materials and production readiness before mass production starts, so there is no finished lot to sample from, and the AQL tables do not really apply.
So knowing what is AQL sampling answers only part of the question. You also have to decide when in the production cycle to apply it. For most importers, the minimum sensible program is a PSI with AQL sampling on every order, plus a DUPRO on first orders, custom products, or high-value runs where catching a problem early is worth the extra inspection day.
Common misunderstandings about AQL sampling
A few myths come up often enough to address directly.
"AQL sampling means some defects are fine." Partly true, and that is the point people miss when they ask what is AQL sampling expecting a guarantee. The method manages risk; it does not eliminate it. You set the limit, and the limit should reflect what your customers will tolerate and what your margins can absorb. For critical defects, the standard practice is an AQL of zero: no tolerance at all.
"A higher AQL number means higher quality." The opposite. AQL 4.0 is more lenient than AQL 2.5, which is more lenient than AQL 1.0. The number is the maximum defective percentage you accept, so a bigger number accepts more defects. This catches a lot of first-time importers out.
"AQL sampling works on any shipment." It works on homogeneous production lots. If your order mixes very different products, colors made with different processes, or goods from multiple factories consolidated into one shipment, one sampling plan across the whole mix can mislead. Split the inspection by product or factory, or at least make sure the sampling covers each variant.
"Passing AQL means the whole shipment is fine." It means the sample passed. A passed lot can still hold defective units outside the sample; that is the statistical trade-off you accepted when you chose sampling over 100% inspection. This is also why your spec sheet and sealed reference sample matter: the inspector can only check against requirements you actually defined.
"Any inspector can do AQL properly." The method is only as good as the sampling discipline behind it. Random selection, correct table lookup, honest defect classification: that is where corners get cut. When you compare inspection providers, ask how they train inspectors on sampling discipline, not just what they charge per day.
Conclusion
What is AQL sampling, in one sentence: a standardized statistical method that lets an inspector check a representative sample of your order and deliver a defensible pass or fail for the whole lot. It rests on the ISO 2859-1 tables, it separates sample size (inspection levels I, II, III) from strictness (the AQL value), and it shows up most often inside a pre-shipment inspection, your last checkpoint before the balance payment.
The practical takeaways are simple. Use Level II unless you have a reason not to. Set your AQL values deliberately instead of accepting defaults blindly, and keep critical defects at zero. Apply the method at the right stage, usually PSI at minimum, and remember the sampling plan is only as good as the spec sheet and reference sample the inspector checks against. Do that, and what is AQL sampling becomes what it is meant to be: a clear, repeatable agreement between you and your supplier about what "good enough" means.
FAQ
### What is AQL sampling in simple terms?
What is AQL sampling, simply put: a method where an inspector checks a randomly selected portion of your order and uses the defect count in that portion to pass or fail the entire lot, following the tables in ISO 2859-1 (ANSI ASQ Z1.4).
### What is the difference between AQL and inspection levels?
AQL values (such as 1.0, 2.5, or 4.0) set how many defects are tolerated. Inspection levels (I, II, III) set how many units get sampled. Level II is the default for most consumer goods; Level I reduces the sample size and Level III increases it.
### Who decides the AQL settings for an inspection?
You do, as the buyer. Your inspection company or agent can recommend settings based on your product category and risk tolerance, but the AQL values and inspection level in the report should reflect your decision, not the factory's preference.
### Does a passed AQL inspection guarantee zero defects?
No. AQL sampling is statistical: a passed lot can still contain defective units that were not in the sample. Critical defects are normally set to an AQL of zero, meaning any critical defect found fails the lot, but for major and minor defects a small percentage is accepted by design.
### Can I use AQL sampling for small orders?
Yes, with limits. On very small lots the sample can end up being a large share of the order, and borderline results carry more uncertainty. For small, high-value orders, some importers prefer 100% inspection or a tightened Level III plan instead.