AQL is not permission for a factory to make defects. It is a way to make a consistent accept-or-reject decision from a sample when inspecting every unit is not practical.
The abbreviation usually means Acceptance Quality Limit. In ISO 2859-1, AQL is defined as the “worst tolerable quality level” for an acceptance-sampling system. The same standard is clear about an important point: setting an AQL does not mean that quality level is desirable. 1
For an importer, the useful question is simpler: What did we agree to inspect, how will we select the sample, what counts as a problem, and what result triggers an agreed response? AQL is part of that answer. It is not the whole quality plan.
What AQL does in a final inspection
With acceptance sampling, the inspector checks a selected sample from a defined lot rather than every item. The result is compared with a sampling plan that specifies a sample size and acceptance/rejection criteria.
ISO 2859-1 defines a sampling plan as the sample size or sizes together with the associated lot-acceptance criteria. 1 QIMA explains the practical table sequence: use the lot size and inspection level to obtain a code letter, then use that letter and the chosen AQL to identify the sample size and accept/reject points. 2
| AQL helps answer | AQL does not answer by itself |
|---|---|
| How many items should be selected under the agreed plan? | What the product should look like, do, or contain |
| How many nonconforming items or findings can occur before the plan rejects the lot? | Whether a product requirement is clearly written |
| Whether the inspected sample meets the agreed acceptance rule | Whether every unit in the whole lot is conforming |
| Whether the same sampling rule is being used across comparable inspections | Whether a laboratory, certification, or regulatory requirement has been met |
| Whether a lot needs an agreed next action under the plan | Who should make a commercial or shipment decision |
AQL is a decision tool. Your specification, approved sample, inspection checklist, and defect definitions tell the inspector what to look for.
Start with a defined lot
A lot is the group of items the sampling decision applies to. ISO 2859-1 defines a lot as a definite amount of product constituted under essentially the same conditions and collected together. 1
This matters because one sampling result should not be stretched across unrelated goods.
| Lot question | Why it matters |
|---|---|
| Is this one SKU, style, color, size, or product family? | Different products may have different requirements and defect risks |
| Was it produced under the same process and current revision? | A material, tool, line, or design change may create a different risk profile |
| Is it the same shipment scope? | An inspection result should be traceable to the goods it represents |
| Are rework goods separated from finished goods? | Mixed status makes a sampling result hard to interpret |
| Is the count credible? | Lot size is an input into many sampling plans |
QualityInspection.org advises considering each different product as a separate lot when an order includes different products. 3 For a real inspection, agree the lot definition in writing with the supplier and inspection provider before samples are selected.
The five pieces of an AQL setup
You do not need to memorize a table to understand the setup. You need to know which inputs the table or provider is using.
| Setup element | Plain-English meaning | Decision to make before inspection |
|---|---|---|
| Lot size | Number of items in the defined inspection lot | Confirm the quantity and whether variants are separate lots |
| Inspection level | The planned sampling scope | Choose a scope that matches product risk, order context, and purpose |
| Sampling plan | The rule that connects the sample size to accept/reject criteria | State the standard, plan type, and any agreed variation |
| Defect or nonconformity class | A category for how serious a failure is | Define the categories for this product rather than relying on generic labels |
| AQL/acceptance rule | The agreed threshold associated with each class | Agree the values and decision rule before the inspector sees the goods |
QualityInspection.org identifies lot size, inspection level, and AQL limits as the key parameters a user needs before using common AQL tables. 3 The table does not make these choices for you.
Inspection level: how much of the lot will be sampled
An inspection level influences the size of the selected sample under the chosen plan. InTouch Quality explains that the general levels GI, GII, and GIII correspond to different sample sizes, with higher levels generally producing larger samples. 4
More sampling may provide more information, but it also requires more inspection effort. It does not remove all uncertainty.
| Consider a broader sampling scope when | A lighter scope may be considered when |
|---|---|
| The product or supplier is new | The product/process is stable and supported by credible, current evidence |
| A defect could be costly, safety-sensitive, or difficult to correct | The product is low-risk and the purpose is limited to a defined check |
| Production has changed or prior issues are unresolved | The order is a repeat under the same controlled product definition |
| The product is highly manual or variable | Product variation is low and the relevant check is straightforward |
| You need more visibility before a decision | You understand what information a smaller sample can and cannot provide |
These are planning considerations, not universal rules. InTouch Quality notes that product risk, supplier history, budget, and inspection purpose affect the suitable scope. 4
Defect classes: agree what the words mean
In day-to-day sourcing, people often use labels such as critical, major, and minor. Those labels are only useful when each label points to a written, product-specific definition.
ISO 2859-1 says that nonconformities may be classified by degree of seriousness and that the number of classes, assignment to a class, and AQL choice should fit the quality requirements of the specific situation. 1
| Class label used by many buyers | A useful way to define it | What not to do |
|---|---|---|
| Critical | A failure that your written requirements treat as unacceptable because of safety, law, a mandatory requirement, or severe use risk | Call something “critical” only because the factory dislikes rework |
| Major | A failure that prevents intended use, creates a material customer problem, or departs significantly from the approved standard | Leave the term undefined and expect the inspector to guess |
| Minor | A limited variance that does not prevent intended use but still departs from the approved standard | Assume every cosmetic issue is minor without a reference or rule |
The words are not a substitute for examples. For each recurring defect, include a description, a reference photo if useful, the product/SKU/version, classification, inspection method, and action rule. Article 35 will cover how to write a defect standard without ambiguity.
AQL is not a simple percentage of the sample
This is where many discussions go wrong.
An AQL value is not a promise that the inspected sample will contain exactly that percentage of problems. It is also not a guarantee that the whole lot has no more than that percentage. Under a sampling plan, the lot size, inspection level, selected table, and AQL setting lead to a sample size and accept/reject points.
QIMA illustrates the two-table logic: one table uses lot size and inspection level to produce a code letter; the next uses the code letter and AQL value to identify a sample size and accept/reject thresholds. 2 QualityInspection.org likewise explains that the tables determine both the sample size and the boundary between acceptance and refusal. 3
Plain-English version: Do not multiply an AQL number by a sample size and invent your own acceptance count. Use the agreed sampling plan or have a qualified inspection provider document the plan it applied.
If an order needs a non-standard plan, make sure the plan, defect definitions, sample-selection method, and supplier agreement are all settled before the inspection. Changing the rules after defects are found damages the usefulness of the result.
AQL needs random selection and clear evidence
Acceptance sampling only makes sense if the selected units give a fair view of the defined lot. A convenient sample pulled from the top carton or a row of pre-selected goods does not provide the same evidence as a documented random selection method.
ISO 2859-1 describes acceptance based on an estimate from a random sample and includes a reference to random sampling and randomization procedures. 1 The inspection record should show what lot was available, how samples were selected, which units were checked, and what was not inspected.
| Evidence to request | Why it helps |
|---|---|
| Lot/quantity statement | Connects the result to the actual goods and planned shipment scope |
| Sampling plan and inspection level | Shows the rule used to select scope and interpret findings |
| Product revision and reference | Shows what the inspector compared the goods against |
| Sample-selection notes | Makes it easier to evaluate whether samples came from across the lot |
| Defect list with photos | Turns a summary result into reviewable evidence |
| Count by agreed class | Connects findings to the acceptance rule |
| Open-item list | Shows what was unavailable, unclear, untested, or outside scope |
Use Pre-Shipment Inspection Checklist for Imported Products to build the broader final-inspection brief. AQL is one section of that brief, not the whole document.
What an AQL pass does not prove
An accepted sample result is useful evidence. It is not a quality guarantee.
ISO 2859-1 says that individual lots with quality as bad as the AQL may be accepted with fairly high probability and that the AQL designation does not suggest a desirable quality level. 1 QualityInspection.org similarly notes that a lot can pass a sampling inspection while still producing an unwelcome surprise for the buyer, and that sampling does not reduce defects in the first place. 3
AQL also does not replace:
- A released product specification and approved reference sample.
- A product-specific inspection checklist.
- Laboratory testing or certification evidence where required.
- Production controls that prevent defects before final inspection.
- A documented response plan for issues found during the visit.
For earlier-stage controls, see Production Quality Checkpoints: A Factory Monitoring Plan. AQL works best as part of a sequence, not as the only quality tool on the order.
AQL and ISO 2859-1: one important nuance
ISO 2859-1 is designed for acceptance sampling across a continuing series of lots, with rules that can move between normal, tightened, reduced, and other inspection arrangements based on quality history. 1
A buyer may use tables derived from this system for a one-off inspection, but should not assume that a single visit delivers every feature of the full standard. If the order requires a formal system, a specialized plan, or a high-consequence decision, get qualified quality and regulatory input before setting the method.
A short AQL setup checklist
Before the inspection is scheduled, make sure you can answer these questions:
- What exactly is the lot, and which products or variants are separate lots?
- Which current product revision, specification, sample, and defect standard will govern the inspection?
- Which sampling plan and inspection level will apply, and who will document them?
- How will samples be selected across the available lot?
- Which defect or nonconformity classes are defined for this product, with examples?
- Which AQL/acceptance rule applies to each class?
- Which product checks sit outside AQL, such as laboratory evidence, document review, or 100% verification of a specific feature?
- Who will review the report, open issues, and evidence under the buyer’s own approval process?
AQL becomes much less mysterious once you see it for what it is: a shared sampling rule. The hard work is still defining the product, the risks, and the evidence you need.