A clothing size chart is a manufacturing document long before it is a web page. The table that decides whether a garment fits on a body is the same table that decides whether a customer keeps it, and it is built from four things: the measurement points you choose, the method used to take each one, the increments between sizes, and the tolerance you allow the factory. Get any of those wrong and production can hit its tech pack exactly while the finished garment still comes back as a return.
This guide is for brand owners sending a first size run to a Bangladesh factory, and for buyers who inherited a chart nobody can explain: which points to record, how grading increments are set for knits and wovens, how to test the chart before bulk, and how to publish it so customers use it.
A Size Chart Is Not a Spec Sheet
Three documents get confused constantly, and the confusion causes most measurement disputes.
A size chart is the table of finished garment measurements, one column per size, one row per measurement point. A spec sheet or tech pack is the wider production document: fabric, construction, stitching, trims, labelling, packing, with the measurement table as one section inside it. A grading chart is the set of increments that generates every size from the base size. Send a factory a size chart without a tolerance column and it will apply its own house tolerance, which may be twice what your fit model needs.
At the counter-sample stage, a measurement out by a centimetre turns into an argument about tolerance, and that argument is already lost if the document never states which numbers are targets and which are limits. For the wider structure these documents sit in, see the guide to what goes into a complete tech pack.
The Measurement Points Every Chart Needs
Pick points by garment type, then define each one in writing. A point defined as "chest" without a stated distance from the armhole seam will be measured three different ways in three different factories.
| Measurement point | How it is taken | Where it matters most |
|---|---|---|
| Chest width | Straight across, 2.5 cm below armhole, seam to seam, garment flat | Knit tops, tees, hoodies |
| Chest circumference | All round at the same depth, garment closed and flat | Wovens, outerwear, anything with a placket |
| Shoulder | Seam to seam across the back, straight line | Structured wovens, jackets |
| Sleeve length | Shoulder seam to cuff edge, or centre back neck to cuff for raglan | Long-sleeve knits, shirting |
| Bicep | Straight across sleeve, 2.5 cm below armhole | Slim-fit sleeves, activewear |
| Cuff opening | Inside edge to inside edge, cuff laid flat and closed | Ribbed cuffs, shirting |
| Body length | Highest point of shoulder to finished hem | Every top, and the first number customers check |
| Waist | All round at the stated depth, or at the narrowest point for a fitted shape | Bottoms, dresses |
| Hip | All round at the widest point, at a stated depth below waist | Bottoms, dresses, tunics |
| Inseam | Crotch seam to hem, inside leg | Trousers, denim, joggers |
| Front rise | Waistband top to crotch seam, front | Denim, chinos |
| Leg opening | Straight across the hem, flat | Bottoms, tapered fits |
Two rules keep this clean. One unit per chart, with the conversion shown rather than mixed into the same column. And one written definition per point, stored with the chart so a new factory is not guessing at method two seasons later. Every point needs a sentence, not a paragraph.
Body Measurements vs Garment Measurements
The single most common error in a published chart is a column headed "chest" that mixes body circumference in one size and garment width in another. They are not interchangeable, and the gap between them is ease.
| Body measurement | Garment measurement | |
|---|---|---|
| What it measures | The wearer | The finished item, laid flat |
| Who owns it | Your fit model or size survey | The factory and the QC team |
| Typically used for | Size recommendation on the product page | Production approval and inspection |
| Relationship | Base for the pattern | Body plus ease, plus fabric behaviour |
| Typical knit top difference | Chest 96 cm body | 100 to 104 cm garment circumference |
Ease is a design decision, not an accident. A fitted jersey top may be designed with 2 to 4 cm of positive ease at the chest; a compression base layer may be cut with negative ease so the fabric stretches to fit. A woven shirt needs enough ease to close over the body without pulling at the placket. Whatever you choose, state it, because the factory grades the garment, not the body.
Size Grading: Setting Increments That Hold
Grading turns one base size into a size run. Increments are chosen per point and per category, and they are rarely uniform across the whole range.
| Category | Typical chest increment | Typical waist increment | Notes |
|---|---|---|---|
| Women's knit top | 2.5 cm | 2.5 cm | Hip often needs 3 cm to avoid a straight cut |
| Men's jersey tee | 3 cm | 3 cm | Length usually grows 1.5 to 2 cm per size |
| Men's woven shirt | 4 cm | 4 cm | Shoulder grows about 1 cm, sleeve 1 cm |
| Women's dress | 2.5 cm | 2.5 cm | Waist-to-hip drop needs its own rule |
| Denim, men's | 3 cm | 3 cm | Rise and thigh grade separately from waist |
| Kids' jersey | 4 cm | 4 cm | Length increments shrink as sizes approach adult |
Those are industry starting points, not laws. The check that matters is at the extremes: lay the smallest and largest size of the graded set next to the base and ask whether the proportions still look like the same garment. A shirt that grades 4 cm at the chest but only 1 cm at the shoulder stops reading as a shirt at XXL. Plus and extended sizes usually need re-grading from a second base block rather than more of the same increment, because body shape changes faster than a linear scale accounts for.
How to Build a Size Chart: Nine Steps
Steps 7 and 8 are where most programmes save money. A chart built from one sample measured by one person encodes that person's habits, not the garment.
Publishing the Chart So It Reduces Returns
The published chart and the factory chart should hold the same numbers, with one addition: a short note on how to measure yourself and which body measurement each size is built for. Two failures show up most often before publication.
The first is copying a competitor's chart. The numbers look plausible, so nobody questions them until returns arrive, and they were built for a different fabric, block and customer. The second is a chart that grades chest and body length only, leaving shoulder, sleeve and bicep to whatever the pattern software defaulted to. Ignoring shrinkage, publishing body measurements while the factory approves garment measurements, and leaving tolerance out of the document all produce the same argument later: a one-centimetre variance with no agreed limit to test it against.
Publish both units, centimetres first and inches second, with consistent decimals. Show garment measurements by default for anything with a deliberate fit, add a body measurement column where the garment is close-fitting, and state what the table cannot: "this style is cut with extra room through the body, take your usual size" prevents more returns than a perfect hip column. Then feed returns data back. If one size accounts for a disproportionate share of exchanges, the increments for that size are usually the problem.
Where Charts Break in Production
Shrinkage is the most predictable problem. Cotton knits commonly move 3 to 5 percent in length after the first wash, and cotton wovens 1 to 3 percent, depending on the yarn, the knit structure and the finishing route. Sanforised or pre-shrunk fabric reduces that, but does not remove it. The fix is procedural: base the chart on after-wash measurements, and state the wash method used to get them.
Measurement method is the second. Two inspectors measuring a chest width with and without the garment relaxed will disagree by half a centimetre on a knit. Write the relaxation time into the inspection standard, then enforce it at the inline and final check. Measurement failures at final inspection are often method failures rather than sewing failures, which is why the inspection standard and the size chart have to be written together; the AQL inspection basics explain how those checks are sampled.
The third is change control. A fabric substitution, a new rib supplier or a different wash house can shift every number in the table. Without a version date on the chart, the factory keeps producing to the old document and nobody notices for two shipments.
Testing the Chart Before Bulk
A pre-production fit session should be small and specific: one sample per size, measured flat, compared line by line against the chart, with any variance logged in centimetres rather than words. Anything outside tolerance goes back for correction before the size set is approved.
Then fit it on people. Three to five wearers spanning the range is enough to expose the failures a table hides: a shoulder that rides, a hem that sits too high when seated, a rise that pulls. Record what they wore underneath, because a fit comment without context is not actionable. The garment sampling process is where these rounds belong, before grading is locked and before fabric is booked.
Only after the fit session passes should the chart be treated as fixed. Changing it after bulk fabric is cut is expensive in a way that changing it on paper is not.
Work With a Factory That Reads the Chart Properly
A size chart only holds if the factory measures the same way you do. TEXVENTURE works as a Bangladesh-based manufacturer with 20+ vetted factories, MOQ from 100 pieces per style, and audits and certifications including ISO 9001, BSCI, OEKO-TEX, SGS, Sedex and GOTS, so a measurement standard agreed at sampling is the standard applied at final inspection. The team handles sourcing through to shipping, and works to the buyer's chart rather than substituting its own block.
If you are launching a range and need a partner who will measure to your tolerances, look at the Trusted Clothing Manufacturer in Bangladesh and ask for a fit sample before you commit to a size run.
Frequently Asked Questions
What measurements should a clothing size chart include?
At minimum, the points that change the fit: chest or bust, waist, hip, shoulder, sleeve length and body length for tops, plus inseam, rise and leg opening for bottoms. Add bicep and cuff for slim sleeves. Each point needs a written definition of where it is taken, because the same word means different things in different factories.
Is a size chart the same as a grading chart?
No. The size chart lists finished measurements for each size. The grading chart is the set of increments used to generate those sizes from a base size. You can publish a size chart without ever showing the grading rules, but the factory needs both documents to produce a consistent run.
How much tolerance should I allow per measurement?
Typical production tolerance is plus or minus 1 cm on stable woven points and plus or minus 1.5 cm on knit points, tightening to 0.5 cm on critical fit points such as chest width and body length. The exact figure matters less than stating it, agreeing it in writing and inspecting against it.
Should the chart show body measurements or garment measurements?
Both, clearly labelled. Customers need body measurements to choose a size; the factory needs garment measurements to approve production. If you can only publish one, publish garment measurements and add a note on how the style is intended to fit.
How many sizes should I grade before the first order?
Grade the full range you intend to sell, even if you only produce the middle sizes first. Charting the extremes exposes proportion problems early, and re-grading after a launch costs more than grading a size you stock later.
Get the Chart Right Before the First Cut
A size chart is cheap to fix on paper and expensive to fix after fabric is cut. Define the points, set the increments, agree the tolerance and test it on people before the size run is approved.
Send your base size, target size range and fit intent through the size chart and fit sample request and the sourcing team will confirm the measurement points, flag anything that will not grade cleanly, and quote the style at MOQ from 100 pieces.