Woman taking picture of new model on fashion show

A fashion model walks the runway. (© oksanazahray - stock.adobe.com)

In a Nutshell

  • Despite more diverse faces and body sizes appearing in fashion media, the typical model body has not changed over 25 years.
  • Non-White models are 4.5 times more likely to be plus-size models than White models, meaning the industry is stacking multiple forms of diversity onto fewer, already-marginalized people.
  • Milan’s numerical runway cutoff coincided with measurable shifts in model body composition; a more flexible French policy showed no comparable pattern.
  • At the most prestigious tier of the industry, the share of extremely thin models was highest, revealing that thinness is most entrenched exactly where influence is greatest.

For years, the fashion industry has patted itself on the back for getting more inclusive. Bigger bodies on runways. More skin tones in magazines. A broader definition of beautiful. But a sweeping new study suggests that much of what the industry calls progress is more symbolic than structural, and now there are nearly 800,000 data points to back that up.

Researchers analyzed close to 800,000 model work records spanning a quarter century of runway shows, magazine covers, advertisements, and editorial shoots. What they found was a stark contradiction: while the fashion world has grown more diverse in certain visible ways, the average model body has remained essentially frozen in time. The so-called “body ideal,” lean, low in body fat, and dramatically different from real American women, hasn’t moved in two decades.

This matters well beyond the pages of a glossy magazine. Decades of research have linked repeated exposure to these narrow body ideals with body dissatisfaction, psychological distress, and disordered eating. Those effects show up across genders and age groups. Beauty standards set by fashion and media shape how ordinary people feel about themselves every day.

Fashion models walk the runway
Despite the push for more diversity in body types and ethnicities, not much has changed on the runway over the past 25 yeras. (© stadelpeter – stock.adobe.com)

Nearly 800,000 Records, One Uncomfortable Truth About the Fashion Body Ideal

Researchers pulled data from two professional modeling industry platforms, covering work records from 2000 through 2024. The final dataset included 793,199 records spanning runway shows, advertisements, magazine covers, and editorial shoots, along with model measurements, visible physical traits, and country of origin. When self-reported identity data wasn’t available (roughly 23% of records), researchers used photo-based software tools to estimate gender and ethnicity. The authors acknowledge this method is imperfect and cannot capture the full complexity of a person’s identity.

To measure body composition without relying on weight, which the modeling industry has historically been reluctant to disclose, the team used a metric called Relative Fat Mass, or RFM. Unlike the more familiar Body Mass Index, RFM uses a person’s height, waist measurement, and sex to estimate body fat levels, and has been shown to be a more accurate measure of actual fat content. This let researchers compare model bodies to those of real American women, using data from a large, long-running federal health survey.

The gap they found was enormous. Female models in the dataset clustered around an average RFM of about 18%, while American women between the ages of 17 and 30 averaged around 38.5% in the federal health data. That’s a gap of roughly 20 percentage points, and it hasn’t narrowed or widened over the entire 25-year study period. Bodies that fashion presents as aspirational are, statistically speaking, almost entirely absent from the general population.

Diversity Grew, But Only at the Edges

So where does the appearance of progress come from? According to the study, published in PNAS, it’s real, but it’s happening at the fringes of the industry rather than at its center.

Over the study period, the range of body sizes represented in fashion media did expand. Standard measurements like waist, hips, and bust showed roughly double the variability by the end of the study compared to the early 2000s. On the surface, that sounds meaningful. But when researchers looked more closely at where that variability was coming from, they found it was almost entirely driven by the addition of a small number of plus-size models, defined by the industry as a U.S. dress size 12 or higher, at the high end of the size range. The middle of the distribution, meaning the most common type of model body, barely changed at all.

“[B]ody-size diversity reflects the inclusion of a few outliers rather than a recentering of the typical model body,” the researchers wrote (p. 4).

Much like a restaurant that adds one extra-large option to the menu while keeping every other item identical, the menu looks more inclusive. The default hasn’t changed.

Meanwhile, diversity in appearance and origin genuinely did grow. Blonde hair and blue eyes became less dominant over time, while darker hair and eye colors gained ground. The share of models from Sub-Saharan Africa, East Asia, and South Asia increased. Eastern European models, who dominated in the early 2000s, declined in representation. These are real, measurable shifts.

Fashion's diversity: progress or illusion? (Infographic)
(Infographic by StudyFinds)

The Fashion Industry’s Diversity Math Has a Problem

Size diversity has been limited to outliers, and it has been disproportionately concentrated on non-White models, a pattern the study describes as “showcase inclusion rather than systemic change.”

Non-White models rose from 13% of the industry in 2011 to 44% in 2024. Plus-size models, meanwhile, grew from nearly zero to about 1% over the same period. But those two trends are far from independent. Starting around 2017, researchers found that non-White models were, on average, 4.5 times more likely to be plus-size models than their White counterparts. That ratio held through 2024.

Researchers describe this pattern as “intersectional concentration,” a situation where the work of representing diversity falls disproportionately on people who are already underrepresented. Rather than broadening the overall model pool, the data suggests body-size diversity is being absorbed by the same individuals who are already carrying racial diversity.

Prestige amplifies the pattern. Researchers ranked fashion brands and magazines based on how connected they were within the industry. At the most prestigious tier, the share of extremely thin models was highest. Plus-size appearances were most common at both the very top and very bottom tiers, but nearly absent in the middle. Mid-tier brands were neither as focused on extreme thinness as elite brands, nor as willing to cast plus-size models as those at the top and bottom of the prestige scale.

Rules With Numbers Work Better Than Rules Without Them

One of the more policy-relevant findings in the study involves a comparison of two regulatory approaches to addressing extremely thin models on the runway.

In 2006, Milan Fashion Week implemented a rule barring female models with a body mass index below 18.5, a specific, numerical threshold. In 2017, France introduced a law requiring models to carry a health certificate from a doctor, but set no specific cutoff for body size. Researchers compared what happened at those cities’ fashion weeks before and after each rule took effect, using other major fashion cities as a reference point.

After Milan’s rule went into effect, average body fat levels among models at that city’s fashion week shifted measurably upward compared to other cities, a pattern consistent with the rule having reduced the presence of the thinnest models on the runway. Those effects held through 2010, the end of the post-intervention tracking window used for that specific comparison.

Paris showed no such pattern after France’s 2017 law. Average body composition at Paris Fashion Week continued to track in line with broader industry trends, with no visible change following the new requirement. When rules are specific and enforceable, they tend to produce results. When they leave room for interpretation, they may be easier to satisfy on paper without changing actual practice.

Beyond fashion magazines and runways, the study’s authors close with a warning about what comes next. Narrow body standards documented over 25 years are now being built into artificial intelligence systems used to generate advertising images. As AI-generated models enter commercial media, researchers caution that those tools risk locking in existing biases rather than correcting them. Without deliberate auditing and transparency standards, the body ideals documented in this study could be embedded into the next generation of visual media at a scale the fashion industry itself has never reached.

Fashion’s diversity era has produced more varied faces and a broader range of origins in its imagery, but the body the industry presents as the standard of beauty remains as narrow, and as distant from ordinary Americans, as it has ever been.


Paper Notes

Limitations

The researchers acknowledge several important limitations. For roughly 23% of records, data on gender and ethnicity was not self-reported and had to be estimated using photo-based software tools. These tools can misclassify individuals and cannot capture the full complexity of personal identity, particularly for transgender, nonbinary, or mixed-ethnicity people. The study also collapses all non-White identities into a single category, which masks meaningful differences across specific groups. Because the modeling industry has historically been reluctant to disclose weight data, the researchers used RFM as a proxy for body fat, a validated measure but not equivalent to a direct clinical measurement. The dataset is drawn from two professional aggregator platforms focused on the internationally oriented fashion circuit, which is itself Western-centric, so findings may not apply to domestic fashion industries in other regions, user-generated social media, or AI-generated content. The two regulatory interventions studied, Milan’s 2006 rule and France’s 2017 law, differ not only in structure but in cultural context and are separated by more than a decade, so any direct comparison should be interpreted carefully. Finally, the population benchmark used to compare model physiques is specific to US women, meaning the documented gap between model bodies and real bodies is strictly a US comparison.

Funding and Disclosures

The authors declare no competing interests. This article is a PNAS Direct Submission and is published as open access under a Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). The authors acknowledged the Santa Fe Institute Complex Systems Summer School, where the study was conceived and the collaboration between the authors began. No specific grant numbers or funding agencies were identified in the source paper.

Publication Details

Paper title: Cultural Evolution of Beauty Standards Authors: Louis Boucherie, Sagar Kumar, Katharina Ledebur, August Lohse, and Karolina Śliwa Journal: Proceedings of the National Academy of Sciences (PNAS) Volume/Issue: Vol. 123, No. 21 Published: May 21, 2026 DOI: 10.1073/pnas.2602380123 Correspondence: [email protected]

About StudyFinds Analysis

Called "brilliant," "fantastic," and "spot on" by scientists and researchers, our acclaimed StudyFinds Analysis articles are created using an exclusive AI-based model with complete human oversight by the StudyFinds Editorial Team. For these articles, we use an unparalleled LLM process across multiple systems to analyze entire journal papers, extract data, and create accurate, accessible content. Our writing and editing team proofreads and polishes each and every article before publishing. With recent studies showing that artificial intelligence can interpret scientific research as well as (or even better) than field experts and specialists, StudyFinds was among the earliest to adopt and test this technology before approving its widespread use on our site. We stand by our practice and continuously update our processes to ensure the very highest level of accuracy. Read our AI Policy (link below) for more information.

Our Editorial Process

StudyFinds publishes digestible, agenda-free, transparent research summaries that are intended to inform the reader as well as stir civil, educated debate. We do not agree nor disagree with any of the studies we post, rather, we encourage our readers to debate the veracity of the findings themselves. All articles published on StudyFinds are vetted by our editors prior to publication and include links back to the source or corresponding journal article, if possible.

Our Editorial Team

Steve Fink

Editor-in-Chief

John Anderer

Associate Editor

Leave a Comment