A size medium from one brand fits you perfectly. A size medium from another brand is two sizes too large. A third brand’s medium doesn’t close. None of this is an accident, a quality control failure, or a design flaw. It’s the entirely predictable outcome of a sizing system architecture that was built a century ago, was never globally standardized, and has been evolving separately in every market, brand, and category since.
Understanding why sizing is broken also explains what would actually be required to fix it — and why the fix is harder than it looks.
Where sizing came from
Modern garment sizing emerged in the early 20th century as the ready-to-wear clothing industry scaled up. The original goal was simple: replace individual tailoring with mass-produced garments that could fit a range of customers without alteration.
The problem was measurement data. Early sizing systems were based on limited surveys, anecdotal knowledge from tailors, and the practical constraints of fabric cutting and pattern making. The United States Department of Agriculture conducted one of the first systematic civilian anthropometric surveys for clothing sizing in 1941 — measuring about 15,000 American women. This survey became the basis for the Mail-Order Association of America sizing standards published in the 1940s.
These standards were immediately flawed. The survey methodology was biased toward certain demographics; the measurements reflected the population of 1940, not subsequent generations; and different brands adopted and modified the standards differently based on their target market.
Since then, American women have grown taller, heavier, and differently proportioned. The sizing standards have barely moved. The gap between the reference body and the actual population body has grown steadily.
Vanity sizing: a rational business decision
The most commonly cited cause of inconsistent sizing is “vanity sizing” — brands deliberately label clothes with smaller size numbers than they historically represented, because customers prefer to wear smaller sizes. A garment labeled “size 10” in 1950 is significantly smaller than one labeled “size 10” today.
This isn’t brands being deceptive. It’s brands making a rational economic decision: customers feel better buying a size 8 than a size 12, even for identical garments, and they return fewer items when they don’t feel the size label is an indictment of their body. The result is size number inflation across the industry.
The problem with vanity sizing isn’t the inflation itself — it’s the inconsistency. If every brand inflated uniformly, you’d simply need to add 2 to every label. But brands have inflated at different rates, in different directions, for different customer demographics, across different time periods. The result is that a size number has essentially no absolute meaning.
Why international sizing doesn’t help
You might expect that international standards would solve this. They don’t — and for a revealing reason.
Several international standards exist: ISO 8559 for clothing construction, EN 13402 for European labeling (which uses actual body measurements rather than size numbers), and various national standards. EN 13402 is the most coherent: it labels garments by the chest, waist, and hip circumferences the garment is designed to fit, rather than a meaningless number.
Why don’t brands use it? Because the number “chest 92” is accurate, but “size S/M” is aspirational. The emotional value of a size label is different from the functional value of a measurement label. Brands that have tried measurement-based labeling report that customers find it confusing — the abstraction of a size number, however meaningless, is more comfortable than the specificity of a measurement.
The fit problem is asymmetric
Here’s a specific cause of inconsistency that’s underappreciated: most garment patterns are designed for one “ideal” body shape and then graded up and down through sizes by scaling patterns proportionally.
The problem: body shape doesn’t scale proportionally. A size 8 person and a size 16 person are not the same shape at different scales. As bodies become larger, some dimensions grow faster than others. Hip-to-waist ratio changes. Bust-to-underbust ratio changes. Shoulder width relative to chest girth changes.
Pattern grading that scales proportionally will fit the central size well and increasingly poorly at the extremes. The same garment in size 8 and size 18 has been designed at two very different points on the body shape spectrum, and the size 18 garment often fits as if it was sized for a uniformly scaled size 8 body — which no real size 18 person has.
The statistical picture
Anthropometric research shows that human body shapes follow a complex multi-dimensional distribution. Height, weight, waist, hip, and shoulder dimensions are correlated but not fully predictable from each other. Two people of identical height and weight may have hip circumferences that differ by 15–20cm. Two people of identical chest measurement may have widely different waist measurements.
This is why measurement-to-size mapping is inherently imprecise. Any size chart that maps a single measurement (or even three measurements) to a size bucket loses information. The real body exists in a high-dimensional space; the size label is a one-dimensional projection.
The information loss shows up as fit failures for the people whose bodies don’t conform to the shape that was assumed during pattern grading. This isn’t a minority. Research on US women’s body shapes found that fewer than 8% of women match the body proportions assumed by standard size chart construction.
What “fit” actually requires
Perfect fit — a garment that conforms to your body’s actual geometry — requires either:
- Custom tailoring from your measurements, or
- A very large number of distinct pattern shapes available to choose from
Most ready-to-wear clothing does neither. The result is that clothing purchase is, for most people, a trial-and-error process: find the brand whose target customer body is closest to yours, learn which size in that brand fits your body, repeat for each new brand.
The digital measurement layer
The most promising structural change in this situation is the ability to systematically measure bodies at scale. If every purchasing decision were accompanied by a dimensional profile — and size charts could be calculated from actual dimensional data rather than from pattern grading from an assumed ideal — the hit rate of first-try purchases would improve.
This is the underlying rationale for body measurement features in e-commerce: not to eliminate the size label, but to map the specific customer’s dimensions to the specific garment’s construction measurements, rather than routing both through a size label that was designed before either existed.
Tools that can predict body dimensions at scale — from height and weight, or from a photo, or from a 3D scan — create the infrastructure for this mapping. The complexity is in the garment data side: building databases of actual garment measurements, not just size labels, is a significant industry undertaking that is still in early stages.
The prognosis
Sizing will not become globally standardized in any near-term future. The economic incentives for brands run counter to standardization, and consumer psychology around size numbers complicates measurement-based approaches.
What’s more likely: continued expansion of digital measurement tools that bypass the size label entirely, matching dimensional data to garment construction data without the size abstraction in the middle. This has been technically feasible for years. The friction has been measurement — getting dimensional data for enough customers to make the matching worthwhile.
As body measurement at scale becomes easier and less expensive, the business case for size label bypass improves. The intermediate step is the sizing recommendation feature that uses body measurements to predict which labeled size fits — not ideal, but better than the size label alone.
The underlying science of human body variation is well understood. The problem has always been measurement at scale, not measurement methodology. That constraint is changing.