How many pieces to inspect

Sample size is set by the order quantity and the acceptable quality limit, not by how much time is available on the day. The table below is the general inspection level used for apparel shipments; if your buyer or your contract names a different level, theirs governs.

Sample size by order quantity
Order quantity	Inspect	Major defects allowed
151 – 500	20 pieces	1
501 – 1,200	32 pieces	2
1,201 – 3,200	50 pieces	3
3,201 – 10,000	80 pieces	5
10,001 – 35,000	125 pieces	7

Take the pieces at intervals through the run rather than from the top of the first carton. The pieces nearest the door are the ones somebody chose.

Measurement points and tolerances

Eight points cover most knit and woven garments. Add whatever is specific to your product — a pocket position, a print placement, a zip length.

Measure flat, relaxed, and record every number
Measurement	Tolerance	How to take it
Chest, 2.5cm below armhole	±1cm	Lay flat, measure across, double it
Body length, HPS to hem	±1cm	From high point shoulder, straight down
Shoulder, seam to seam	±0.5cm	Across the back, point to point
Sleeve length, shoulder to cuff	±0.7cm	Centre of shoulder seam to cuff edge
Cuff opening, relaxed	±0.5cm	Measured flat, not stretched
Neck width, seam to seam	±0.5cm	Inside the collar, laid flat
Hem width, straight across	±1cm	Flat, no tension
Weight per piece	±5%	One piece on a 0.01g scale

The ten checks

These are ordered by how much they cost when they are missed, not by how easy they are to make. The first two find most substitutions.

What to check, and what counts as a failure
Check	What counts as a fail
Fabric weight	Under the approved weight by more than 5%. Weight is the fastest way to find a substitution, and the easiest thing in a specification to misstate.
Colour, against the approved standard	Any visible difference under daylight against the signed standard. Not under factory lighting, which flatters everything.
Seam integrity	Any seam that opens under a firm hand pull at the seam allowance. Test five pieces across the run, not the one you were shown.
Stitch density	Fewer than 8 stitches per inch on a main seam. Count it on the fabric, do not count it off a photograph.
Buttons and hardware	Any button that pulls out under a 10kg static load, or any attachment that is loose at the thread. Shake the garment and listen.
Symmetry	Visible mismatch between left and right at collar points, pocket placement or sleeve length. Fold the garment in half and look.
Measurement against the approved size chart	Any measurement outside tolerance in the table above, on more than the sample count allows.
Labels and care instructions	Wrong content, wrong position, or missing fibre composition. This one is a customs problem as well as a quality problem.
Needles and metal contamination	Anything that a metal detector flags. Non-negotiable: this is the one defect that can injure a customer.
Packing	Polybag gauge, carton count, polybag warnings, or the packing list not matching what is physically in the carton.

Recording the result

A defect with no number beside it does not support a claim. Record, for every check: how many pieces were inspected, how many failed, and the specific measurement or photograph for each failure. Then write one of three decisions and date it — approve, rework, or reject — and send the report the same day.

Two habits make the difference between an inspection and a walk round a factory. Photograph every failure with something for scale in frame, so nobody can dispute what was measured. And keep the packed samples you inspected, in case the goods that arrive are not the goods you saw.

What this checklist is not

It is not a substitute for a third-party inspection on a large or high-value order, and it is not a certification. Nobody who inspects a shipment certifies a factory — that is a different discipline with a different qualification, and conflating the two is how "we inspected it" starts meaning "we guarantee it".