Why cotton shrinks, and what pre-shrunk does not promise

1 October 2026 9 min readApolyth

A cotton t-shirt that comes out of its first wash shorter has usually not been damaged. It has gone back towards a shape it was pulled out of before it was sold. Most of what gets called shrinkage is cloth relaxing, and the two useful questions are how much stretch was still in it, and whether anyone took it out before it reached you.

The stretch goes in at the mill

SANFOR, which licenses the Sanforized trademark, describes yarn and cloth as under continuous tension through spinning, weaving, bleaching, dyeing and finishing, stretching to suit it. On a knitting machine the loop itself is formed under tension, and the mechanisms that draw the fabric along add more. Wet processing most often stretches a knit in length and pulls it in at the width. A Cotton Incorporated technical bulletin puts it plainly: "a knit fabric acts like an accordion". Squeezing water out of wet knitted fabric, it says, commonly stretches it 10 per cent or more.

Dried in that state, the cloth stays stretched, held there, SANFOR explains, by friction within the fabric.

What water does

In the wash, water and detergent act as a lubricant, and with the machine's movement the fibres relax back towards their unstretched length.

Cotton also swells. Measuring the same fibres wet and dry under a microscope in 1950, Anna Moore, Lorraine Scott, Ines deGruy and Mary Rollins found the cross-section grew by 21 to 34 per cent of its dry area, whatever the variety or maturity. Flax and a viscose sample swelled about twice as much, and nylon showed no cross-sectional swelling. A 2024 review finds that cellulose fibres typically swell several times more across than along their length, so a wet yarn gets fatter rather than longer.

As the yarns swell, the crimp in a knitted loop, its waviness, increases. The loops pull towards a rounder shape, their lowest-energy state, and in Cotton Incorporated's words "the rounding of the loop results in a shortening of the loop". Synthetics neither swell nor deswell like this, and can be heat-set; cotton cannot.

What the dryer does

Drying adds shrinkage of its own, the Cotton Incorporated bulletin explains: as the fibres give up water and deswell, the structure "shrinks upon itself". Its tests of cotton knits tumbled without heat found the same pattern in each: an immediate shrink as tension let go, little change while surface water evaporated, then rapid shrinkage once the moisture in the fibres fell below about 20 per cent. Tumbling without tension, the bulletin says, is "a form of mechanical compression". Line or flat drying, it notes, releases mostly the quick tension-release part, while tumble drying realises "all available residual shrinkage", the shrinkage still left in the cloth.

Heat usually takes the blame. In one Cotton Incorporated test, a cotton interlock, a double knit that looks the same on both faces, was tumble dried at 38°C, 66°C and 94°C and shrank by identical amounts. That is one fabric, and it concerns the dryer. For the wash, a 2000 study of knitted fabrics put through repeated wash and tumble-dry cycles recorded the highest shrinkage in loosely knitted cotton at the highest laundering temperature. Neither settles how much heat contributes at home.

One wash and dry may remove around 90 per cent of the available shrinkage, Cotton Incorporated estimates, though full relaxation can take several cycles.

How shrinkage is measured

A shrinkage figure means little without its test. ISO 6330 sets out 35 domestic washing procedures across three kinds of reference machine and six drying procedures from line drying to tumble drying, and notes that machine type, detergent and dryer "can affect test results". ISO 3759 and ISO 5077 cover measuring the specimens and working out the change. The American equivalents are AATCC TM135 for fabrics and TM150 for garments, each with four washing temperatures, three agitation cycles and four drying procedures. AATCC states that its standard procedures "represent, but may not exactly replicate, current consumer practices".

As their published scopes describe them, these are methods, not pass marks. The acceptable figure is agreed between a buyer and a mill, and Cotton Incorporated warns that some shrinkage specifications "may not be attainable at all".

Compacting and sanforising

Both shrink the cloth mechanically at the mill, so less is left to happen later. Sanford L. Cluett of Troy, New York, filed the patent in 1930. The cloth is moistened until it is "internally mobile", pressed together along its surface to increase the "yarn undulation or crinkle" left by weaving, and then set by drying. SANFOR describes today's machine: damp cloth is fed onto a rubber belt whose surface has been stretched by squeezing, and as the belt recovers it carries the cloth with it, shortening the warp, the lengthwise yarns. Knits go through compactors that push the loops shorter and rounder.

In one Cotton Incorporated example, a cotton single jersey, the plain knit with a smooth face, had 12.1 per cent residual shrinkage in length and 15.9 per cent in width straight off the knitting machine, measured over five home launderings. Dyed and dried relaxed, with softener only, it was 7.3 by 3.5; dyed, dried and compacted, it was 4.5 by 6.0: less shrinkage in length than relaxed drying alone, but more in width. Compaction can also come undone: a 1999 Cotton Incorporated bulletin on wet processing warns that too much silicone softener can stop it working, or release it during cutting and sewing, and the 2004 bulletin adds that cutting and sewing can add shrinkage of their own.

Sanforized is a licensed trademark with a standard. A woven fabric carrying it must not shrink or grow more than 1 per cent in either direction in an ISO 6330 wash test, with up to 1.5 per cent more allowed after tumble drying and making up. A knit under the Sanfor-Knit label is allowed 5 per cent, tested at 60°C with tumble drying.

What pre-shrunk can promise

In May 1938, Commissioner R. E. Freer of the US Federal Trade Commission described draft rules for woven cotton that would forbid "non-shrinkable" outright, and allow "pre-shrunk" only with "a specific statement as to the limits within which residual shrinkage may be expected to occur". The rules were approved that June. A year later Freer set out what they meant: "when a term such as pre-shrunk is used without qualification, it means that there is no residual shrinkage". The FTC's linen rules of 1941 took the same line, allowing an unqualified "Preshrunk" only where no residual shrinkage remained, and otherwise giving model wording with a blank to fill:

"Preshrunk (or shrunk)-will not shrink more than ----%."

In 1979 the Commission rescinded all of its trade practice rules. Title 16 of the US Code of Federal Regulations, where the FTC's rules sit, no longer contains the word "preshrunk". The FTC's Care Labeling Rule requires a maker to have a reasonable basis for its care instructions, such as reliable evidence that a garment "was not harmed when cleaned reasonably often according to the instructions on the label", without saying how much shrinkage counts as harm. In the EU and Great Britain, the textile labelling regulation covers fibre names and composition and does not deal with shrinkage.

What remains is general consumer law. In 1978 the FTC held that misrepresenting "the amount by which a fabric will shrink when it is washed" is an unfair or deceptive practice, in a case where an importer's invoices claimed residual shrinkage of 1 to 1.5 per cent for fabric that tested at about 5 to 6 per cent. In the UK, goods sold by description must match the description, and giving false or misleading information about a product is a misleading action.

None of the current rules cited here defines pre-shrunk. In the FTC's 1938 and 1941 rules, the bare word meant no residual shrinkage at all. Whether a regulator or a court would read it that strictly today, these sources do not test. The 1941 rule left a blank where the number goes. The blank is still the part that matters.

Sources

Everything above that is a claim about the world rather than about us is checkable here. If we have got something wrong, we would like to know.