AQL Sample Size for Lot 1,000 — GI Inspection at AQL 2.5
What this means
For a production lot of 1,000 garments at GI inspection level with an AQL of 2.5 (major (standard) defects), ISO 2859-1 requires the buyer's QC inspector to pull 20 garments at random from across the lot. If they find 3 or fewer defects, the entire lot is accepted. 4 or more defects means the lot is rejected — the factory must then sort 100% of the lot, repair defective pieces, and re-present for inspection.
Inspection level note: Reduced inspection — used for low-risk products or trusted suppliers.
AQL tier: Industry-standard major defect AQL — broken stitch, open seam, wrong measurement.
Accept / reject decision table
| Defects found in sample | Decision |
|---|---|
| 0 defects | ACCEPT |
| 1 defect | ACCEPT |
| 2 defects | ACCEPT |
| 3 defects | ACCEPT |
| 4 defects | REJECT |
| 5 defects | REJECT |
Worked example — your shipment
Suppose your factory has produced 1,000 garments for a buyer who specified GI / AQL 2.5. The buyer's inspector arrives, opens 5-8 cartons at random, and pulls 20 garments across them.
If the inspector finds 3 defects in those 20 garments → lot ACCEPTED, container ships.
If the inspector finds 4 defects → lot REJECTED. You sort 100% of 1,000 pieces, repair, and re-present in 2-3 days.
How this combination compares
This is the GI / AQL 2.5 combination — the major (standard) sampling plan for GI inspection. If your buyer's PO sheet specified different parameters, here are the adjacent plans:
FAQ for Lot 1,000 · GI · AQL 2.5
What if I find exactly 3 defects?
3 defects = ACCEPT. The accept number is the maximum number of defects you can have in the sample without rejecting. Exactly at the limit is still acceptable.
What if I find 4 defects?
4 defects = REJECT. The reject number is the lowest defect count that fails the lot. There is no "borderline" — anything ≥ 4 triggers a reject.
Can I argue an inspection where I found 4 defects?
You cannot argue the math — the accept/reject thresholds come from the ISO 2859 statistical tables. What you can argue is the classification of each defect. If the inspector marked something as a Major defect that your buyer's spec sheet calls Minor, push back with the spec sheet in writing.
What's the underlying acceptance probability at AQL 2.5?
ISO 2859-1 plans are calibrated so that lots actually at the AQL value get accepted ~95% of the time. So a lot truly at 2.5% defect rate has a 95% chance of being accepted by this plan. Lots at 2× the AQL get rejected most of the time.
Use this in your own QC software
The same logic is published as a free MIT-licensed package:
$ npm install garment-aql-calculator $ pip install garment-aql-calculator // Node.js const { calculateAQL } = require('garment-aql-calculator'); calculateAQL({ lotSize: 1000, inspectionLevel: 'GI', aql: 2.5 }); // → { codeLetter: 'F', sampleSize: 20, acceptNumber: 3, rejectNumber: 4 }