How much science is there on facial attractiveness? We counted every paper in PubMed

Someone tells you in a comment thread that “studies show” your jaw shape matters. You ask the obvious follow-up: how many studies, exactly? The answer is often delivered with total confidence and no count, no query, and no way to check.

That bothered us, so we ran the searches. This page gives you the numbers, the exact strings behind them, and the distinction that matters most: a paper about a measurement is not automatically evidence for the attractiveness claim attached to it.

Key numbers

• 963 papers: our PubMed search for `"facial attractiveness"[Title/Abstract]` returned 963 papers. Open the live search.
• 39 papers: our PubMed search for `"canthal tilt"[Title/Abstract]` returned 39 papers. Open the live search.
• 894 papers: our PubMed search for `"gonial angle"[Title/Abstract]` returned 894 papers. Open the live search.
• 55 papers: our PubMed search for `"golden ratio"[Title/Abstract] AND (face[Title/Abstract] OR facial[Title/Abstract])` returned 55 papers. Open the live search.
• 7 papers: our PubMed search for `looksmaxxing[All Fields] OR looksmax[All Fields]` returned 7 papers. Open the live search.

Table of contents

• How much research is there on facial attractiveness?
• How we counted
• The vocabulary gap: how the forum terms compare
• Why gonial angle has 894 papers and canthal tilt has 39
• What the field actually studies
• What a paper count cannot tell you
• The bottom line
• Sources

How much research is there on facial attractiveness?

There are 963 PubMed papers matching our exact query, `"facial attractiveness"[Title/Abstract]`. The indexed literature was sparse in the 1970s and 1980s, expanded through the 1990s and 2000s, then accelerated in the 2010s and early 2020s.

The same query returned 3 papers in 1990, 4 in 2000, 29 in 2010, 56 in 2020, 70 in 2022, and 71 in 2025. The 2026 publication year is incomplete, so we do not treat it as comparable to a full year.

This is not a story about a field that has “proved” one facial formula. It is a broad research area becoming more active while online vocabulary becomes more specific than the evidence usually is.

Photo: Armin Rimoldi on Pexels

How we counted

We used the public PubMed E-utilities API from the National Library of Medicine on 2026-09-04. The headline query, run exactly as written, was `"facial attractiveness"[Title/Abstract]`. For decade counts, we ran the same query with `datetype=pdat` and `mindate=maxdate=<year>`; comparison terms were restricted to Title/Abstract except where noted. The API documentation is in the NCBI E-utilities guide.

PubMed indexes biomedical and life-science literature, not every psychology or economics journal. These counts are a floor, not a census: they compare topics inside one consistent index.

The vocabulary gap: how the forum terms compare

The forum vocabulary is not a map of the research base. Our position is simple: the vocabulary grew faster than the evidence. We are not saying canthal tilt is fake; 39 indexed papers simply cannot, by themselves, justify the hierarchy a subculture has built around it.

Here is the comparison, using our 2026-09-04 PubMed searches:

The table shows a mismatch between broad academic labels and narrow forum labels. Read our looksmaxxing pseudoscience analysis, facial symmetry explainer, and golden-ratio fact check as interpretation, not as substitutes for reading papers.

Why gonial angle has 894 papers and canthal tilt has 39

Gonial angle has 894 papers because it is an ordinary measurement in orthodontics, oral surgery, and forensic anthropology. Most are not about attractiveness; they concern jaw growth, surgery planning, or skeletal identification.

Canthal tilt has 39 papers because it is a specialized label, not because a low count proves the feature irrelevant. A count may cover clinical contexts rather than controlled tests of the forum claim.

Our cross-query makes the point: `"facial symmetry"[Title/Abstract]` returned 1,177 papers, while `"facial symmetry"[Title/Abstract] AND attractiveness[Title/Abstract]` returned 61. A large measurement literature can shrink when you add the outcome you care about.

That is the measurement-to-claim check: take the term, add the claimed outcome, and inspect what remains. It is better than treating a large search result as proof.

A caveat: Title/Abstract searches miss papers that use different terminology, and adding a word can exclude relevant work. The check improves discipline; it does not produce a perfect bibliography.

What the field actually studies

The indexed field spends more effort on broad, testable patterns and applied questions than on the forum’s favorite facial geometry. Our comparison searches returned 102 papers for averageness of faces, 76 for waist-to-hip ratio and attractiveness, 271 for first impressions of faces, and 31 for machine learning and facial attractiveness.

Averageness asks how typical structure relates to judgments; waist-to-hip ratio concerns body shape, not a male facial prescription; first-impression research asks what people infer quickly; machine-learning work asks whether models can predict or classify judgments, not explain human attraction.

Langlois and colleagues’ review contains eleven meta-analyses and reports strong agreement among raters within and across cultures, while emphasizing whole-face judgments rather than a feature checklist. Rhodes reviews averageness, symmetry, and sexual dimorphism, while noting that explanations remain contested. That is more careful than “one angle determines your appeal.”

For men specifically, our query `"facial attractiveness"[Title/Abstract] AND (male OR men OR masculine)[Title/Abstract]` returned 367 papers. That is meaningful coverage, but not a personal forecast: study population, task, and image set still matter.

What a paper count cannot tell you

A paper count cannot tell you quality, replication, effect size, or whether a result applies to your face in your setting. It cannot tell you whether papers agree, whether participants or images recur, or whether the search term best names the construct.

PubMed is not all of science, and 963 papers matching `"facial attractiveness"[Title/Abstract]` are not 963 replications. A term can be well studied under another name; it can also appear without being a paper’s central question.

One context check is worth making without turning this into a diagnosis. Our search for `"body dysmorphic disorder"[Title/Abstract]` returned 1,792 papers, of which 207 also mention male, men, or boys. The literature on appearance-related distress is larger than the literature on the particular forum claim many people are distressed about. That should encourage better questions, not more panic.

If you are using research to inspect your own appearance, keep the frame healthy: your face is not a problem set that becomes more solvable the longer you stare at it. Practical presentation changes can help, but anxiety is not evidence that a measurement deserves more authority.

Real World Appeal adds the missing axis: perceived first-impression response, rather than pretending a clinical measurement is a complete answer. You can run a test, see the result before deciding anything, and there is no paywall on your result. The caveat is important: an AI report is a structured perspective, not a lab result or a promise about how every person will react.

Photo: SHVETS production on Pexels

The bottom line

There are 963 PubMed papers matching our exact `"facial attractiveness"[Title/Abstract]` query, but confident forum terms do not necessarily have the deepest evidence. Canthal tilt appears in 39 papers, mewing or tongue posture in 41, and looksmaxxing in 7 under the queries shown above. Gonial angle’s 894 papers mostly measure a clinical or forensic construct, not attractiveness.

The useful conclusion is not that research is fake. It is that the vocabulary grew faster than the evidence. Re-run the searches, add the outcome word, inspect the population, and refuse to confuse an indexed mention with support for a personal rule. Then test your own presentation and treat the result as information, not a verdict.

Sources

• Langlois, J. H., Kalakanis, L., Rubenstein, A. J., Larson, A., Hallam, M., & Smoot, M. (2000). “Maxims or myths of beauty? A meta-analytic and theoretical review.” Psychological Bulletin, 126(3), 390-423. PubMed record.
• Rhodes, G. (2006). “The evolutionary psychology of facial beauty.” Annual Review of Psychology, 57, 199-226. DOI record.
• National Library of Medicine. E-utilities documentation, used for the reproducible PubMed searches run on 2026-09-04.
• Real World Appeal. Headline facial-attractiveness search, canthal-tilt search, gonial-angle search, golden-ratio search, and looksmaxxing search, all run on 2026-09-04.