S SoftMaxx ← All articles
Measurement · Face shape

Facial width-to-height ratio (fWHR) and bizygomatic width: the measurement, the claim, and the replications

The most-cited number in face-shape psychology is also the most thoroughly re-tested. The balanced read, with effect sizes.

Published 2026-09-07 · ~8 minute read · Editorial standards

Bizygomatic width is the distance between the two zygion points — the most lateral points of the cheekbones. The facial width-to-height ratio (fWHR) divides that width by upper-face height, the vertical distance from the upper lip to the brow (or, in many studies, to the upper eyelid). A wider, shorter face gives a higher ratio.

fWHR became famous after a 2008 paper linked it to aggression in men. Since then the two claims attached to it — that men have higher ratios than women, and that a higher ratio predicts dominant or aggressive behaviour — have both shrunk sharply under larger, better-controlled replication. It remains a legitimate measurement; it is a weak predictor, and we say so because most face-rating content does not.

1. How it is measured — and why the ruler matters

The founding behavioural study defined fWHR as the "distance between the lip and brow (height of upper face)" divided into the distance between "the left and right zygion (bizygomatic width)" (Carré and McCormick, 2008, Proceedings of the Royal Society B). Many later studies, including the largest lifespan dataset, use upper-eyelid to lip for height and temple-to-temple for width instead (Summersby and colleagues, 2022, Royal Society Open Science).

Those are different ratios. A value from one tool cannot be compared with a value from another unless both landmark sets match — which is the first thing to check before reading meaning into a number an app gives you.

Bizygomatic width on its own is a standard anthropometric measure (zy–zy). The reference norms are the international study of 1,470 healthy 18–30-year-olds by Farkas and colleagues (2005, Journal of Craniofacial Surgery) and the US occupational survey by Zhuang and colleagues (2010, Annals of Occupational Hygiene). We have not reproduced their per-population millimetre tables here because we could not open them to check; one recent single-population sample reported 138.4 ± 6.9 mm in men and 132.8 ± 5.3 mm in women (Choudhary and colleagues, 2026, Indian medical students) — an illustration of scale, not a norm.

2. The claim: the 2008 paper

The idea that fWHR is a sexually dimorphic trait came from skull measurements: in 121 southern African crania, upper-face height was similar in males and females while facial breadth was larger in males (Weston, Friday and Liò, 2007, PLoS ONE). Carré and McCormick then reported that in men — not women — fWHR predicted reactive aggression in the lab (r = 0.38, 37 men), penalty minutes in 21 varsity hockey players (r = 0.54), and penalty minutes in 112 NHL players (r = 0.30).

Note the samples: 37, 21 and 112 men. Effects that size from samples that size are exactly the kind that shrink when re-tested, and that is what happened.

3. What the replications found

The sex difference mostly vanished. In four large adult samples, three showed no sex difference in fWHR at all, and the fourth showed men lower than women until body-mass index was controlled — while three other validated facial measures showed large sex differences in every sample (Lefevre and colleagues, 2012, Evolution and Human Behavior). Across 17,607 passport photographs, mean fWHR was 2.19 ± 0.17 in women and 2.18 ± 0.18 in men — a sex effect of η²p = 0.001 — with men exceeding women only around ages 27–33 and women exceeding men after 48 (Summersby and colleagues, 2022).

The behaviour link shrank to small. Two 2015 meta-analyses, one by fWHR sceptics and one by its proponents, converged: fWHR–aggression in men r = 0.11 (18 studies, 4,141 men; Haselhuhn, Ormiston and Wong, PLoS ONE) and fWHR–threat behaviour r = 0.16 (22 studies, 4,603 men; Geniole and colleagues, PLoS ONE). Lab studies showed roughly double the effect of field data (r = 0.21 vs 0.09) — the usual small-sample inflation signature. Geniole's analysis also found fWHR correlates with BMI at r = 0.31, so a wide face is, in part, a heavier face.

The largest datasets found nothing. In 137,163 people across 55 psychometric scales, fWHR was "not substantially linked with any" self-reported behavioural tendency (Kosinski, 2017, Psychological Science). Among more than a thousand executives rated by colleagues who knew them, there was "very little evidence of fWHR predicting antisocial tendencies" (Wang and colleagues, 2019, Psychological Science).

4. Perception is the part that survives — for now

What does replicate is that observers react to the ratio: pooled across 37 studies, higher fWHR faces were rated more threatening (r = 0.46) and more masculine (r = 0.35) (Geniole and colleagues, 2015). That is a fact about raters, not about the person being rated — the 2019 executive study framed its residual effect as biased perception rather than behaviour.

Even the perception finding is less stable than it looks. In a preregistered study of 11,481 raters across 11 world regions judging 120 faces, men's fWHR showed no reliable association with any of 13 trait judgements; women's fWHR did show reliable negative associations with several, including dominance, trustworthiness and attractiveness (Durkee and Ayers, 2021, Evolution and Human Behavior). Which sex the effect appears in depends on the study.

5. Reading your own ratio

If you want the number: on a true frontal photograph with no head tilt, mark the two zygion points (the widest points of the cheekbones, usually level with the lower orbital rim), the upper lip's top edge, and either the brow line or the upper eyelid crease — and say which. Divide width by height. Typical adult values cluster near 1.8–2.2 depending on the landmark set used.

Then treat it as a description of face shape and nothing more. A high ratio tells you your face is relatively wide for its upper height. The evidence does not license anything further about your temperament, and at the population level it barely distinguishes men from women.

6. What fWHR does not tell you

Related measurements: facial thirds and fifths, gonial angle, the 11 facial metrics. How SoftMaxx measures width and height is on the methodology page.

7. Sources

8. Common questions

What is a normal facial width-to-height ratio?

Across 17,607 adult passport photos, the mean was about 2.2 (women 2.19 ± 0.17, men 2.18 ± 0.18) using lip-to-eyelid height. Studies using lip-to-brow height give lower values. The number depends on the landmarks.

Do men have a higher fWHR than women?

Barely, if at all. The meta-analytic sex difference is d = 0.11; three of four large samples found none; the largest lifespan dataset found an effect size of η²p = 0.001, reversing after age 48.

Does fWHR predict aggression?

Weakly at best. Meta-analyses put the correlation in men at r ≈ 0.11–0.16 (1–3% of variance), and the largest studies of self-reported and colleague-reported behaviour found essentially nothing.

What is bizygomatic width?

The distance between the two zygion points, the most lateral points of the cheekbones. It is the numerator of fWHR and a standard anthropometric measurement; the reference norms are Farkas (2005) and Zhuang (2010).

Is a high fWHR attractive?

There is no consistent evidence that it is. Raters associate higher ratios with perceived dominance and masculinity in some studies, but a large preregistered cross-cultural study found no reliable trait associations for men's fWHR and negative ones, including for attractiveness, for women's.

Keep reading