AQL 2.5 Explained: A Practical Guide for Indian Garment Exporters

AQL 2.5 means you accept a lot if the long-run average percentage of defective units is 2.5% or lower. For a garment shipment, you draw a small sample using ISO 2859-1 tables, count defective pieces, then accept or reject based on the plan. Single-sampling, normal severity, general inspection level II is the most common variant.

If you export garments to international brands, you've heard "AQL 2.5" in compliance manuals, buyer audit reports and shipment-release emails. Most factory teams know it's about sampling for quality, fewer know how the numbers actually work, and almost nobody on the floor can answer what changes when a buyer specifies "AQL 1.5" or "AQL 4.0" instead.

This guide explains AQL 2.5 the way a QC supervisor or merchandiser would actually use it. We cover the ISO 2859-1 tables, the sample-size logic, the accept/reject decision, defect classification, common mistakes — with one worked example end-to-end for a 5,000-piece export order.

What does AQL actually stand for?

AQL stands for Acceptance Quality Limit (previously called Acceptable Quality Level — the name changed in the 1999 revision of ISO 2859-1, but everyone still says both). The technical definition is the worst tolerable process average: if your factory's long-run defective rate is at or below the AQL number, the sampling plan is designed to accept your lots with high probability — about 95% acceptance for most AQL values when the lot quality matches the AQL.

A few clarifications that matter on the floor:

• AQL is not a defect tolerance for the inspected lot. It does not mean "we allow 2.5% defectives in this carton." It's a long-run process target. A specific lot at exactly 2.5% defectives has about a 95% chance of being accepted by an AQL 2.5 plan — but lots at higher defective rates have decreasing chances.
• AQL is a number you and your buyer agreed on. It comes from the buyer's compliance manual or the purchase contract. The buyer may specify different AQLs for critical, major and minor defects.
• The sampling plan converts the AQL into actual sample-size and accept/reject numbers depending on the lot size. That's what the ISO 2859-1 tables give you.

Why AQL 2.5 specifically?

AQL 2.5 is the most common acceptance limit for major defects in garment, footwear and consumer-products manufacturing. Brands settled on it as a workable industrial standard — strict enough that factories can't get away with sloppy production, loose enough that 100% inspection isn't necessary.

You'll see different AQLs in the same audit, applied to different defect categories. A common buyer manual setup:

• Critical defects: AQL 0 (any critical defect rejects the lot)
• Major defects: AQL 2.5
• Minor defects: AQL 4.0

Critical defects are typically safety issues — broken needles, sharp objects, mislabelled flammability. Major defects are functional or visible workmanship problems — open seams, broken stitches, wrong measurements outside tolerance, mis-matched labels. Minor defects are cosmetic — slight shading, loose threads, minor uneven stitching.

The ISO 2859-1 sampling plan

The AQL number on its own doesn't tell you how many pieces to inspect. For that, you need ISO 2859-1 (also published as ANSI/ASQ Z1.4 in the US — the tables are nearly identical). The procedure for a single sampling plan is straightforward once you know it.

Step 1 — pick the inspection level

ISO 2859-1 has three general inspection levels (I, II, III) plus four special levels (S-1 to S-4) for destructive testing or when sample reduction matters. Most garment shipments use General Inspection Level II — it's the default unless the buyer specifies otherwise.

• General I: smaller samples, lower discrimination — used when production has demonstrated consistent quality
• General II: standard — used by default
• General III: larger samples, higher discrimination — used when quality history is uncertain or buyer wants extra rigour

Step 2 — look up the sample-size code letter

Find your lot size on the ISO 2859-1 Table I (the code letter table). The lot size range maps to a letter from A (smallest) to R (largest).

For General Inspection Level II:

Step 3 — look up sample size + accept/reject numbers

With the code letter, go to ISO 2859-1 Table II-A (the single sampling plan for normal inspection). The intersection of code letter + AQL gives you three numbers: sample size, accept number (Ac), and reject number (Re).

For AQL 2.5, single sampling, normal severity, the relevant rows are:

These numbers are stable across versions of ISO 2859-1 since the 1999 revision. If your audit report shows different values, check the inspection level and severity — those are the most common reasons for surprises.

A worked example

You're shipping 5,000 pieces of a knit polo to a US buyer. The buyer's manual specifies AQL 2.5 for major defects, single sampling, normal severity, General Inspection Level II.

1. Lot size 5,000 → Code letter L (the 3,201–10,000 row)
2. Code L, AQL 2.5 → Sample size 200, Ac 10, Re 11
3. Pull 200 finished pieces at random from across the lot (ideally proportionally from different cartons, not just the top layer of one)
4. Inspect each for major defects per your defect catalogue
5. Count the major defects across the 200 sample pieces
6. If you find 10 or fewer major defects: accept the lot. The shipment is cleared for dispatch
7. If you find 11 or more major defects: reject the lot. The lot is held for 100% re-inspection and rework

A few practical notes that compliance manuals usually skip:

• One piece can contribute multiple defects. A piece with both an open seam and a missing label has two majors, not one. AQL counts defects, not defective pieces, in the original formulation.
• Many buyers count defective pieces, not defects. Read the manual carefully. If they say "we count defective units," then a piece with both an open seam and a wrong label still counts once. The accept/reject numbers stay the same; what's counted changes.
• Pull samples at random from across the lot. "Random" means using a sampling table or random-number generator, not picking the visible top layer. Buyer auditors look for sampling discipline.

Single sampling vs double sampling

Single sampling — what we covered above — is the most common plan. You inspect one sample, decide accept or reject.

Double sampling is an alternative the standard provides. You inspect a smaller first sample. If defects are very low, you accept; if defects are very high, you reject; if defects are in between, you inspect a second sample and combine the results.

Double sampling can be more efficient for clearly good or clearly bad lots — you decide faster with fewer pieces inspected. The downside is operational complexity: your QC team needs to handle "inspect again" without losing track. Most factories stick with single sampling unless the buyer specifies otherwise.

Severity: normal, tightened, reduced

ISO 2859-1 has three severities: normal, tightened and reduced. Most lots default to normal severity. The standard's switching rules tell you when to escalate or relax:

• Switch from normal to tightened when two out of any five consecutive lots have been rejected on original inspection. Tightened severity uses tighter accept/reject numbers — fewer defects allowed at the same AQL.
• Switch from tightened back to normal when five consecutive lots are accepted on original tightened inspection.
• Switch from normal to reduced (when permitted by the buyer) requires sustained quality history — typically ten consecutive accepted lots, low total defects, and stable production.

In garment manufacturing, normal severity is the default. Buyers occasionally switch you to tightened after a rejection; reduced inspection is rare.

Defect classification — the most-skipped step

The AQL plan tells you how many defects trigger rejection. It doesn't tell you what counts as a defect — that comes from your defect catalogue. A good defect catalogue is the single biggest determinant of how AQL audits land.

Three categories are standard across most buyer compliance manuals:

1. Critical — defects that compromise safety or regulatory compliance. Examples: broken needles, sharp metal points, mislabelled fibre content where regulation matters, mislabelled care instructions. AQL 0 in most buyer manuals — any critical rejects the lot.
2. Major — defects that affect product function, appearance or saleability. Examples: open seams 5mm or longer, broken stitches, missing buttons, wrong measurements outside ±1cm tolerance on key points of measure (POMs), wrong or missing labels, fabric holes, stains larger than a standard reference circle. AQL 2.5 in most buyer manuals.
3. Minor — defects that are cosmetic and don't affect function. Examples: minor uneven stitching, slight shading variance, loose threads under a length threshold, small fabric flaws within tolerance. AQL 4.0 in most buyer manuals.

The factory's job is to translate the buyer's compliance manual into a clear, photo-illustrated defect catalogue your QC inspectors use during sampling. Without that catalogue, two inspectors looking at the same defect will reach different conclusions, and your AQL numbers will drift.

Common mistakes factories make

After watching dozens of audits across Indian garment factories, the same mistakes show up.

1. Inspecting only from accessible cartons. Auditors and good internal QC pull samples proportionally from across the lot, not from the cartons stacked by the door. Concentrated sampling produces unrepresentative numbers and gets flagged.
2. Mixing severities silently. A defect catalogue with three major-defect descriptions next to two minor-defect descriptions where the difference is "buyer-specific" — without writing down which buyer says what — creates ambiguity. QC inspectors guess, audit outcomes vary, and management can't trace why.
3. Skipping documentation when a lot is borderline. A lot with 10 major defects at code letter L (Ac 10) is at the absolute boundary of acceptance. Document the inspection conditions, the inspector, the timestamp, the carton sources — because if the buyer challenges that lot later, your audit log is your defence.
4. Re-inspecting only the rejected pieces. When a lot is rejected and you go to 100% inspection for rework, the standard expects you to inspect everything, not just the pieces that failed the sample. Skipping this creates ghosts — defects that were in the unsampled pieces and travel through to shipment.
5. Treating AQL as an OQ — outgoing quality target. Some factories set AQL 2.5 as a goal: "our defective rate is 2.5% or lower." But AQL 2.5 isn't an aspiration; it's the worst quality your buyer will tolerate before lots start getting rejected. If your factory averages 2.5% defective, half your lots will fail audit eventually. Aim well below the AQL — most disciplined garment factories run major defective rates of 0.5% to 1.0% to leave headroom.

What buyer compliance manuals get right (and wrong)

Buyer compliance manuals across major US and EU garment brands are largely aligned with ISO 2859-1. The variation is mostly in defect classification — what counts as a major versus a minor varies by buyer, fabric, garment construction, end-use. A polo for a department store may have different POM tolerances than a polo for a premium label even if both are AQL 2.5 normal severity.

The other variation is in supplemental tests — colourfastness, dimensional stability after wash, fibre composition, child labour audits, packaging compliance, e-tag readability. These aren't AQL plans; they're separate tests with their own pass/fail criteria. AQL covers the workmanship sample, not the lab tests.

If you're an exporter handling multiple buyers, the most useful thing your factory can do is build a single internal defect catalogue with buyer-specific overrides documented separately. That way your inspectors apply the same baseline judgement every time, with documented deltas where a specific buyer says something different.

Recording AQL audits — what good factories do

For each AQL audit you run, capture:

1. The lot identifier (order number, style, colour, size split)
2. The lot size
3. The inspection level and severity used
4. The AQL number applied (often three numbers — critical / major / minor)
5. The code letter, sample size, and accept/reject numbers from the table
6. The actual sample drawn (with carton sources documented)
7. The defects found, classified, with photos for any disputed call
8. The accept/reject decision, with the inspector's name and timestamp
9. Any rework or re-inspection actions taken if rejected

For Indian garment exporters running multiple buyers and multiple orders concurrently, capturing this in a structured system — not a notebook — is what turns AQL from a buzzword into a working quality system. The defect logs become the historical record that lets you spot which operators, lines or styles produce the most defects and improve operations before the next buyer audit.

Further reading

If AQL is one piece of the QC + operations puzzle, here's what pairs with it:

• Garment factory workflows — cutting plans, BOM versions, T&A milestones, AQL audits in one operational stack.
• WhatsApp ERP for Indian Factories — the floor-data-capture layer that feeds QC entries.
• Best ERP for MSME Manufacturers in India (2026) — comparison of eight ERPs by manufacturing depth and QC support.
• Quality features inside UdyogFlow — three-stage QC, AQL templates, photo defect logs.

A note on UdyogFlow

UdyogFlow ships AQL 2.5 and 4.0 templates with sample-size auto-pick from ISO 2859-1 tables, photo-attached defect logs and rework loops. For factories that want their AQL audits to be structured, photo-documented and buyer-shareable as PDF without rebuilding the system every time a buyer updates their compliance manual, that's the value. See how Quality fits into the broader manufacturing workflow or explore garment-specific workflows.

Sources

• ISO 2859-1:1999 — Sampling procedures for inspection by attributes — Part 1: Sampling schemes indexed by acceptance quality limit (AQL) for lot-by-lot inspection. The canonical international standard for attributes sampling. Published by ISO; widely adopted in apparel, electronics and consumer-goods manufacturing.
• ANSI/ASQ Z1.4-2003 (R2018) — the American National Standard equivalent, maintained by ASQ. Tables align with ISO 2859-1 for most practical purposes.
• Bureau of Indian Standards IS 2500 — Indian Standard for attribute sampling, harmonised with ISO 2859.
• AAFA (American Apparel & Footwear Association) — buyer-supplier compliance manuals for member brands; widely referenced in Indian export shipments.
• The Apparel Export Promotion Council (AEPC) and Federation of Indian Export Organisations (FIEO) publish industry guidelines aligned with international AQL sampling practice.
• Buyer technical manuals — most large brands publish a compliance handbook to suppliers with their specific AQL setup, defect classification and severity escalation rules. These are private documents and vary by brand.