Fabric Testing: The Four Tests That Stop Expensive Mistakes

A founder places a hoodie order with us.

The fabric looks right on the swatch: good hand, right weight, and color approved.

Bulk runs and garments ship.

Then the emails start. The black has gone charcoal at the shoulders.

The body has shrunk half a size after one wash.

The surface has fuzzed up under the backpack straps.

None of it showed on the swatch, because a swatch that sits on a desk never gets washed, rubbed, or worn.

That’s the whole problem fabric testing solves.

Fabric rarely fails when you’re looking at it.

It fails after a wash, after friction, after a season of use.

Testing is how you make the fabric fail in a lab, on your schedule, before you’ve committed the money instead of on a customer after you have.

This guide gives you two things to remember.

First, the four failure modes, the four ways fabric betrays a garment, and the one test that catches each.

Second, the three gates are the three points in a production run where you actually run those tests.

Get both right, and you’ve closed off the great majority of “the fabric was wrong” disasters.

This is a general production guide, not a substitute for a lab’s own protocol, but it’s the framework we run every order through on our own floor in Sialkot.

The Four Failure Modes:

01. The core tests: what actually fails, and how to catch it

You could run dozens of textile tests.

For a streetwear brand shipping tees, hoodies, and fleece, four of them catch nearly every failure a customer will ever complain about.

Each maps to one way fabric fails in the real world.

That’s the map.

Now, each one in turn:

What it is, how the lab runs it, and what a fail looks like on a finished garment.

Shrinkage: (dimensional change)

The lab cuts a specimen, marks reference points a set distance apart, then washes and dries it under a defined home-laundering cycle, usually three cycles.

Re-measure, and the change is reported as a percentage in length and width.

Negative is shrinkage; positive is growth (yes, some knits grow).

The standard behind it is AATCC TM135 for fabric or AATCC TM150 for a made-up garment, with ISO 6330 (the wash procedure) and ISO 5077 (the measurement) as the international equivalents.

As a rough industry benchmark, woven fabric is often held to no more than about 3% change and knits to about 5%, though plenty of brands run tighter.

What a fail looks like: a hoodie that’s a full size shorter after the first wash.

A tee whose graphic is distorted because the body shrank around the print.

Shrinkage is the single most common “the fit changed” complaint, and it’s invisible on an unwashed swatch.

Colorfastness: (to washing and to rubbing)

Two tests live under this heading, and streetwear needs both.

Colorfastness to washing (AATCC TM 61) is an accelerated wash; a specimen is tumbled with detergent and steel balls in a canister for around 45 minutes, simulating roughly five home washes.

It’s graded on two gray scales: color change (did the shade shift?).

Staining (did color transfer onto a white multifiber strip washed alongside it?).

Both run 5 (no change) down to 1 (severe).

Colorfastness to crocking/rubbing (AATCC 8, or the ISO 105-X12 equivalent) checks whether color rubs off dry or wet.

This is the one that catches dark garments marking pale bags, sofas, and skin.

The ISO family ISO 105-C06 (wash), ISO 105-X12 (crocking), and ISO 105-B02 (light) mirror the AATCC set for buyers who work with ISO.

What a fail looks like: a black tee that’s washed-out grey by month two. A new hoodie that dyes a white tee in the same wash.

A print that ghosts onto a backpack. Dark reactive-dyed cotton and pigment-dyed goods are the usual offenders.

Pilling:

Pilling is measured by abrading a specimen and grading the surface against photographic standards, from 5 (no pilling) down to 1 (severe).

The two common routes are the random tumble method (ASTM D3512), where specimens tumble in a chamber, and the ISO 12945 family: the pilling box (ISO 12945-1, good at provoking loose knits) and the modified Martindale (ISO 12945-2).

Many retailers set a pass around grade 3-4.

Because washing loosens fabrics, a serious protocol is often graded after a wash, not just off the roll.

What a fail looks like: the fuzzy little balls that form on a fleece hoodie exactly where a backpack, a seatbelt, or an arm rubs.

Soft-handed cottons and brushed-back fleece feel great, and the worst pill is that trade-off is real, and testing tells you where a given fabric lands.

Strength:

For woven fabric, strength usually means tensile and tear strength.

For the knits that make up most streetwear, the relevant test is bursting strength: a specimen is clamped over a diaphragm that inflates until the fabric ruptures, and the pressure at burst is recorded (in kPa or psi).

The methods are ASTM D3786 (diaphragm/Mullen) and ISO 13938-1 (hydraulic).

ASTM’s own text describes D3786 as satisfactory for acceptance testing of commercial fabric shipments, which is exactly the decision you’re making.

What a fail looks like: a seam or a stressed panel that splits at an elbow, a shoulder, or the belly of an oversized tee.

Low burst strength is often a symptom of an overshaved, too-light, or over-processed knit that felt luxuriously soft on the swatch.

This page explains how the tests are run and gated.

For what a fabric spec sheet is (GSM, composition, construction), see the fabric weight and GSM guide, and for turning tests into an inspection sign-off, see the quality control and AQL guide.

The Three Gates:

02. When and why to test before bulk:

Knowing the four tests is half the job.

The other half is running them at the right moment.

Test too late, and you’ve already paid for bad fabric.

The reasoning is simple, and it’s the reason this whole page exists:

A failure caught on a swatch costs a swatch.

The same failure caught after bulk costs the order.

Every step downstream multiplies the cost of a defect.

A color problem spotted on a lab dip costs a re-dip.

Spotted on a production sample, it costs a sample round and a week to produce.

Spotted after bulk fabric is knitted and dyed, it costs the fabric often thousands of dollars of committed dye lot.

Spotted after garments are cut and sewn, it costs the whole order plus the shipping plus the refunds. So you don’t test once.

You test at three gates, each cheaper than the one after it.

The Bulk Gate is the one founders skip, and it’s the most expensive to skip.

The sample fabric passing through tells you the quality is achievable.

It doesn’t tell you whether the bulk lot achieved it. Dye lots vary. Knit runs vary.

The only way to know your actual bulk is good is to test your actual bulk.

Where this sits in your timeline matters: the Desk Gate runs alongside choosing and planning your production timeline, and the Sample Gate rides on top of your pre-production sample stage.

Testing isn’t a separate project; it’s a checkpoint you attach to steps you’re already doing.

Which is also why it’s usually resolved during the sampling process rather than as an extra round.

The Report Read:

03. How to read a test report without getting bluffed:

A test report is only as good as your ability to read it.

A weak supplier can hand you a page that says “PASS” in bold and means very little.

Here’s how to read one properly and how to spot a report that’s hiding something.

The five things a real report shows

A trustworthy report always names five things.

If any are missing, treat the result as unproven:

1. The method and its year: “AATCC TM135-2022,” not just “shrinkage test.” Standards get revised; the year tells you which one ran.
   

   ---

2. The conditions: wash temperature, number of cycles, and dry method. “3 cycles, 40°C, tumble” is a different test from “1 cycle, cold, flat.”
   

   ---

3. The specimen: how many pieces, from where. One lucky specimen isn’t a result.
4. The actual number is the measured value, not a verdict. “2.4% length / 1.1% width,” or “grayscale 4-5,” not a bare “pass.”
   

   ---

5. The number vs a stated standard, the pass/fail line it was judged against, so you can check the verdict yourself.

Now the grades themselves, because the scales point in different directions and it’s easy to misread:

The Pass Line:

04. Setting your pass/fail standards:

A test result is a number. A standard is the line you judge it against.

Without a pre-agreed line, every result becomes a negotiation with your supplier after the fact, which you’ll usually lose because the fabric’s already made.

Set the line before you test, in writing, in your tech pack or fabric spec.

Here’s the honest part: the right line depends on your product, your price point, and your market’s requirements.

What follows are common industry starting points, not universal law. Use them as a floor to reason from, then tighten where your product demands it.

The other half of your standard: how much you inspect

Fabric-property standards decide whether the material is right.

An AQL (Acceptable Quality Limit) decides how many finished units you inspect and how many defects you’ll accept before rejecting the lot.

It’s the sampling side of quality, set by ANSI/ASQ Z1.4 (the international equivalent is ISO 2859-1), and it’s the standard your final inspection is built on.

AQL is a full topic of its own, with the sampling tables, the defect classes, and the inspection levels, and it’s covered in the QC and AQL guide.

For fabric testing, the point to hold is this: your fabric thresholds and your AQL are two halves of one standard.

One says the material is right; the other says the batch is clean enough to ship.

Skip either, and “we tested it” means less than it sounds.

The Gate Grid:

05. Building testing into your QC plan:

Everything so far becomes useful the moment it stops being a decision you make each time and becomes a gate every order passes through automatically.

That’s a QC plan: the four tests, run at the three gates, judged against your written standards, with a defined action on a fail.

The whole system collapses into one grid.

What gets tested at which gate:

The plan needs four decisions written down once so nobody improvises mid-order:

• What do you test the rows above, trimmed to your product? A pigment-dyed heavyweight tee tests crocking and shrinkage; a brushed fleece tests pilling.
  

  ---

• Against what line the thresholds from Chapter 4, in the spec, agreed before bulk.
  

  ---

• Who pays and who runs it: factory in-house, third-party lab, or split. Agree to this in the contract, not the crisis.
  

  ---

• What happens on a fail: hold, retest a fresh specimen, re-process (re-wash/re-dye), or reject the lot. A fail with no defined action is just a bad feeling.

Written down, this is a one-page document your factory and your inspector both work from.

It turns “I hope the fabric’s okay” into “The fabric cleared three gates against a standard we set,” which is exactly the difference between a brand that eats returns and one that doesn’t.

Four Failure Modes tell you what to test. Three Gates tell you when. The QC plan makes it happen every time, without you deciding again.

Quick glossary: