New: Body Time Machine and Outfit Try-On are live. All measurement tools stay free. Explore tools →
Physique

Face Attractiveness: What Research Says and What You Can Actually Change

Which facial ratios research links with perceived attractiveness - symmetry, averageness, facial thirds and fifths, and the golden ratio - how much they matter, how to measure yours, and what is and is not trainable.

5 min read · Updated September 28, 2026 · Not medical advice

"How attractive is my face?" is searched even more often than the body version of the question, and the honest answer is the same: no single number captures it. What research has repeatedly measured are a handful of facial ratios connected with a face being perceived as attractive. This guide covers what those are, how strong the evidence behind each one really is, and which parts you can actually change.

The short version

  • Symmetry has the most consistent evidence: faces with smaller left-right differences tend to be rated more attractive, though the effect is modest.
  • Averageness is a well-replicated finding: faces closer to the population average (not extreme in any one feature) tend to be rated more attractive than faces further from it.
  • Facial thirds and fifths are a classical proportion canon, still widely taught in illustration and cited in some clinical literature, but modern measurement studies find most real faces deviate from it - it is a rough guide, not a rule.
  • The golden ratio is the most popularly cited and the most scientifically contested of the four.
  • Taste varies. Every one of these findings describes an average tendency across many people, not what any one person will prefer.

Symmetry

Fluctuating asymmetry - small, random left-right differences that are not the mirror-image "handedness" every face has - has been studied since the 1990s as a possible signal of developmental stability. Researchers including David Perrett's group at St Andrews found that digitally increasing facial symmetry in photographs raised attractiveness ratings, and the effect replicates reasonably well across studies, though it is smaller than popular science articles often suggest.

In a photo, apparent asymmetry is very sensitive to how it was taken. A tilted head, an off-centre camera, or a smile on one side can create differences that are not really there in person. Before reading much into a low symmetry score, retake the photo face-on with a neutral or even expression. You can test yours with the Face Attractiveness Test, and compare it with body symmetry using the Body Symmetry Test.

Averageness

In a widely cited 1990 study, Judith Langlois and Lori Roggman digitally averaged multiple faces together and found the composite images were consistently rated as more attractive than most of the individual faces that made them up - a finding often summarised as "attractive faces are only average." The effect has been replicated many times since, though later work shows extremely average is not the ceiling: symmetry and a few other cues can make a face more attractive than a pure average.

Averageness is not something a proportion calculator can score directly, since it depends on comparison against a large population sample rather than a fixed target ratio. It is included here because it is one of the best-evidenced findings in the field, and it is a useful reminder that "unusual" facial proportions are not automatically a downside - they are simply further from an average that itself varies by population.

Facial thirds and fifths (the classical canon)

Renaissance artists and, later, clinicians described an idealised face divided into three equal horizontal sections (hairline to brow, brow to nose base, nose base to chin) and five equal vertical fifths (each about one eye-width wide). It is still taught in illustration and referenced in some orthodontic and plastic-surgery literature as a starting point for facial analysis.

Measurement studies of real faces, notably work by Leslie Farkas and colleagues using anthropometric data from thousands of subjects, find that most faces deviate from the classical canon by a meaningful margin, and that the canon predicts attractiveness only weakly, if at all, once symmetry and averageness are accounted for. Use it as a descriptive benchmark rather than a target to hit.

The golden ratio face

The golden ratio (phi, about 1.618) is the most widely repeated claim about facial beauty, popularised by cosmetic-surgery marketing and "beauty calculator" apps that measure face length against face width. The evidence is mixed: a few small studies report a weak association between golden-ratio-like proportions and attractiveness ratings, while others find no reliable effect once symmetry and averageness are controlled for. Treat closeness to phi as a playful benchmark, the same way this site treats it for the body in the golden ratio body guide.

What you can change - and how fast

Signal Trainable? Best levers Typical timescale
Apparent symmetry in a photo Mostly photo setup Face-on angle, level camera, neutral expression, even lighting Immediate (retake the photo)
Jaw and cheekbone taper No (bone structure) Facial hair shape, hairstyle change how the line reads Immediate (styling)
Facial thirds / forehead-to-jaw balance No (bone and hairline) Hairstyle that exposes or covers the forehead Immediate (styling)
Skin, grooming Yes Skincare routine, sleep, hydration, sun protection Weeks to months
Underlying bone structure No - -

Because the true geometry cannot change from a tip, most of the realistic "improvement" in a face score comes from fixing the photo (a face-on angle with even light removes most of the symmetry penalty) rather than the face itself.

How to get a fair measurement

  1. Face the camera straight on, eyes level with the lens, with your whole face and hairline in frame.
  2. Use even, soft lighting on both sides of the face - avoid a single strong light source or backlight, which exaggerates asymmetry.
  3. Keep a neutral or relaxed expression, hair off the forehead, and remove glasses or sunglasses.
  4. Shoot from about arm's length or further with a normal lens; a very close-up shot (common on phone selfies) distorts the nose and lower face.

The Face Attractiveness Test does the measuring in your browser, breaks the score into four parts and tells you which one would move it most. For the body-proportion side of the same question, see the Body Attractiveness Test and the Body Attractiveness Guide. Your photo is never uploaded.

What this does not measure

Real-world facial attractiveness includes skin condition, expression, grooming, styling, how a face moves when someone talks or smiles, and every bit of context a photo cannot capture. Research on first impressions consistently finds the face carries a large share of a snap judgment, but once people actually interact, personality, warmth and familiarity change perception more than any measurement ever could. Use a face score the way you would use any measurement: something to understand and, where it is genuinely trainable (mostly the photo itself), to improve - not a rating of your worth.

Frequently asked questions

What facial features are most linked to attractiveness?
Two findings show up again and again: symmetry (how closely the left and right sides match) and averageness (how close a face sits to the population mean). Facial thirds and fifths - a classical proportion canon - and a gentle jaw-to-cheekbone taper are also commonly cited, though the evidence for them is weaker and more debated than for symmetry and averageness.
Is the golden ratio real for faces?
It is a popular claim, not a settled scientific finding. Some studies report faces closer to golden-ratio proportions are rated slightly more attractive; others find no reliable effect once averageness and symmetry are controlled for. Treat it as a fun benchmark rather than a target.
Can I change my face attractiveness score?
The underlying bone structure will not change from a tip. What does change the number: retaking the photo face-on with even lighting and a neutral expression (this fixes most apparent asymmetry), and grooming or hairstyle choices that change how the jawline, forehead and cheekbones read in a photo.
Is facial attractiveness the same everywhere?
Symmetry and averageness preferences are found fairly consistently across cultures. Preferences for specific proportions, such as how narrow a jaw or how full a lip should be, vary more between cultures, eras and individuals, so treat published ratios as tendencies, not rules.
How much does the face matter compared with the body?
For first impressions, most research finds the face carries a larger share of perceived attractiveness than the body. A face score and a body score measure different things - see the Body Attractiveness Guide for the physique side.