How do I know if my face is symmetrical?

Take a front-facing photo, then mirror it and compare the two versions. The differences that jump out when you switch between them are your largest asymmetries. For a measured answer, upload the photo here and the tool reports how far five paired regions miss their mirror image, in proportion to the width of your face, so you can see which region is furthest off rather than forming a general impression.

Why is my face not symmetrical, and is that normal?

It is normal to the point of being universal. Paired features develop with small independent differences, and the face carries more of them than most parts of the body because it has so many paired features sitting next to each other where they can be compared. Daily habits add more on top: most people chew more on one side, sleep more on one side, and hold a resting expression that pulls one corner of the mouth further than the other. So the useful question is not whether your face is asymmetrical but which regions are furthest off, and whether the gap is one anyone else would ever register. Usually it is not, because faces are read as wholes, at conversational distance, in motion. What is not in that ordinary category is asymmetry that appears suddenly, affects one whole side, or comes with drooping, numbness or pain, which is a question for a doctor rather than a web tool.

Is facial asymmetry noticeable to other people?

Far less than it is to you. You see your face at close range, in a mirror, while studying it, which is the condition under which small differences are most visible. Other people see it at conversational distance, in motion, unmirrored, and read it as a whole. The mirrored composites on this page are useful for finding your asymmetry but actively misleading about how visible it is, because they double every difference.

What is facial symmetry?

It is how closely the two halves of a face mirror each other across its midline. In a measurement context it is specific: you define an axis, take pairs of points that correspond across it, reflect one side onto the other, and record how far each pair misses. Facial symmetry is therefore a matter of degree rather than something a face either has or lacks.

What is a good symmetry score?

There is no agreed answer, and be wary of a tool that gives you one without qualification: symmetry scores are not a shared unit, so two tools will hand the same photo two different numbers and neither converts into the other.

Does a perfectly symmetrical face exist?

No. Perfect symmetry would mean every paired point on your face reflects onto its partner exactly, and measured finely enough that is not true of any face that has ever been measured. A test that returns 100 percent is reporting its own resolution rather than a perfect face. This is also why a high score does not make the two mirrored halves look identical: the composites double whatever difference exists, so a small one is still clearly visible. The related idea, that a more symmetrical face is a more attractive one, is a separate claim with a much messier evidence base than the confident version usually quoted. Symmetry is one input among many, and not the dominant one.

What is the difference between a symmetrical and an asymmetrical face?

It is a difference of degree, not of kind. Every face is asymmetrical to some extent, so the labels describe the ends of a range. A face called symmetrical has paired features that miss their mirror positions by a little; one called asymmetrical misses by enough to be visible in ordinary viewing. Where between those the line sits is a judgment, not a measurement.

Does facial symmetry make you more attractive?

It is one input, and a weaker one than the popular version of the claim suggests. Symmetry is easy to measure, which is part of why it gets quoted so often. Proportion, feature harmony and plenty of things no geometric tool touches at all carry more of the impression, and none of it is measured against a single scale.

Which parts of the face affect symmetry most?

In measurements from photographs, the jaw and the nose usually carry the largest deviations, the jaw because it holds much of the underlying structural difference and the nose because deviation there is common and sits at the center of the face where it is easy to see. Brow and mouth pairs move most with expression, which makes them the least stable across photos of the same person. This tool reports all five separately for exactly that reason.

Face Symmetry Test, And What That Percentage Is Really Measuring

Upload one photo and this face symmetry checker gives you three things: one number for how far your two halves miss each other, the two mirror halves built from your own face, and a region-by-region reading of where the differences actually sit. The last two are the ones that hold up.

Free while in beta, no app needed. The measuring runs on your own device.

Analyze your face

Deterministic geometry, not an AI opinion. Runs entirely in your browser.

🔒 Your photo is read locally and scored on your own device. Nothing is uploaded to score it.

Drag a photo here, or click to choose one

JPEG or PNG. Front-facing, well-lit, one face per photo.

Or try an example

Ready.

For a good result

Face the camera directly, front-on
Good, even lighting on your face
Only your face in frame, not cropped at the chin or forehead
A sharp, in-focus photo

What you get back

One photo in, three things out. This is a real run on the example face, not a mock-up — the figures below are what the tool actually returned for it.

Two faces, both built entirely out of one person. Neither is how anyone sees her — the composites double every difference — but the gap between them is her asymmetry, made visible without a single number.

Average deviation

1.2%

of face width

By region, largest first

Jaw largest: 2.4%
Mouth corners: 2.1%
Nostril edges: 0.9%
Brows: 0.5%
Outer eye corners: 0.1%

The ordering is the part that holds up. That her jaw misses by more than twenty times what her outer eye corners do is a fact about her face; whether 1.2% counts as good is a question nobody can answer yet, and this page says so further down.

Run it on your own photo

How symmetry gets measured

How to check face symmetry without a tool

You do not need software for a usable answer. Three methods, in order of how much they tell you.

1.Flip the photo

Take a front-facing photo and look at it, then mirror it horizontally and look again. You have spent your life seeing your face in a mirror, so the flipped version is the one closer to what you are used to and the unflipped one is what everyone else sees. Whatever jumps out when you switch between them is your largest asymmetry.

2.Draw the midline

On the same photo, draw a straight line through the middle of the bridge of your nose, tilted to match the line of your eyes rather than straight up and down. Then check each pair against it in turn: brows, outer eye corners, nostrils, mouth corners, jaw. Working pair by pair stops you from forming an overall impression before you have looked at anything specific.

3.Cover one half

In a mirror, cover one half of your face with your hand, look, then swap. This is the low-tech version of the mirrored composites and it isolates each side the same way, without exaggerating the difference the way a doubled composite does.

Both lines are on the same photo of the same head, tilted about sixteen degrees. The midline follows the line of your eyes, not the edge of the frame, so it is the one the five pairs are actually measured against. The line on the right was computed from the eye centres rather than drawn by eye.

Doing this by hand and running the tool are worth combining. The hand methods tell you what is noticeable to a person looking at you, which is the question most people are actually asking. The tool tells you how large each difference is in proportion to your face, which the eye is bad at judging.

How a facial symmetry test measures a face

Every automatic face symmetry test does the same three things. It places landmark points on your face. It works out where the midline of your face is. Then it takes pairs of points that should mirror each other across that midline, measures how far each pair misses by, and averages the misses into one number.

The second step is where tools quietly differ, and it matters more than the first. The cheap way to find a midline is to assume the photo is upright and treat the vertical center of the face box as the axis. Do that and every head tilt gets scored as facial asymmetry, because a tilted face genuinely is asymmetrical with respect to a vertical line. Tilt your head five degrees between two photos and your score moves, with nothing about your face having changed.

This tool builds the axis from your face instead of from the photo frame. It takes the center of each eye, draws the line between them, and uses the perpendicular bisector of that line as the midline. The axis leans when your head leans, so the measurement follows your face rather than the camera. Then each left-side point is reflected across that axis and compared against its right-side partner.

The five pairs that get compared

The midline is built from the eye centers, so it tilts with the head. The dashed circle at pair 4 is where the left mouth corner lands once it is reflected across the midline; the short line to the right corner is the distance being measured.

1.Brow

The inner-to-mid brow point on each side. Brow height is the pair most often moved by expression rather than by structure.

2.Outer eye corners

Measured at the outer corner rather than the eye center, because the eye centers are what define the midline in the first place and would score a perfect zero by construction.

3.Nostril edges

The outer edge of each nostril. Nose deviation is the pair people most often notice in their own photos.

4.Mouth corners

The pair most sensitive to a half-smile, which is why a neutral expression gives a reading you can repeat.

5.Jaw

A point on each side of the jaw below the ear. This pair carries most of the underlying bone asymmetry, and also most of the error when the head is turned.

Each miss is divided by the width of your face before it gets used, so the result is a proportion rather than a pixel count. A ten-pixel offset means something very different on a face that fills the frame than on one shot from across a room, and dividing it out makes the two comparable.

The five proportions are then averaged and mapped onto a 0 to 100 score. That last mapping is the weakest link on this page, and the section on scores below explains why.

Your face symmetry checker, left and right

The score is an abstraction. The mirror halves are not. Take the left side of your face, reflect it across the midline, and join it to itself, and you get a face made of two left sides. Do the same on the right and you get a second face. Both are built entirely out of you, and most people find they look like two different relatives.

That gap is the clearest picture of your own asymmetry you can get, and it needs no calibration, no reference population and no threshold to interpret. You are not comparing yourself against anyone. You are comparing you against you.

It is also worth knowing what the composites are not. They exaggerate. Joining a half to its own reflection doubles every deviation, so a face that reads as slightly uneven in a photo can read as clearly uneven in the composites. This is the same effect behind the mirrored-face clips that circulate as a filter or a meme. It makes small differences visible, which is the point, but it is not a preview of how anyone else sees you.

Reading your result

What is a good face symmetry score?

This is the most common question in this category and it deserves a straight answer: nobody can currently tell you, and the tools that give you a confident one are not measuring something you can check.

A symmetry percentage is not a shared unit. Two tools that pick different landmark pairs, weight them differently, normalize by a different width or draw the midline a different way will hand the same photo two different numbers, and both can be internally correct. A 92 from one site and an 85 from another are not eight points apart. They are on separate scales with no conversion between them.

People often arrive here holding a number from somewhere else and wanting it decoded. The honest decoding is that the number described one run of one algorithm on one photo. It is not a property of your face the way your height is a property of your body.

The score on this page has the same limitation and we would rather say so. The measured deviations are real, and they are reproducible: give this tool the same photo and it returns the same proportions every time. The step that turns those proportions into a score out of 100 uses a constant we picked before having real photos to set it from. Until enough have come through, treat the score as a rough band, not a rank.

What the number is good for anyway

Comparing your own regions against each other

If your nose deviation is triple your brow deviation, that ordering is a genuine finding about your face and does not depend on the calibration at all.

Comparing two of your own photos taken the same way

Same lighting, same distance, same neutral expression, and the difference between the two readings means something.

Sanity-checking what you think you see

People routinely believe one specific feature is badly off and find their largest measured deviation somewhere else entirely.

Deviation, not score: everything on this scale was measured directly, unlike the step that turns it into a mark out of 100. The band is twenty-nine official portrait photographs run through this tool, covering where the middle half of them landed. Twenty-nine faces are not a population, which is why the page still will not hand you a percentile.

What it is not good for is placing yourself against other people. This page does not report a percentile and will not until there is a real sample behind one. A tool that tells you that you are more symmetrical than 78 percent of people is quoting a distribution it has not shown you.

Why the same face gets two different symmetry scores

Run any face symmetry test twice on two photos of the same person and the results will differ, often by more than the difference between two people. This is the single most useful thing to understand about the category, because it explains nearly every confusing result.

A face is a three-dimensional object and a photo is a flat projection of it. Turn your head a few degrees and the far side of your face is compressed by the projection, which moves the jaw and nostril points on that side inward. The measurement is correct about the photo. The photo is simply no longer a fair view of the face.

What moves the reading, and what to do about it

Head rotation, the big one

A small turn toward one shoulder shortens one half of the face in the image, and the jaw pair takes the hit first. Face the camera straight on, with both ears equally visible.

Head tilt

The midline is built from your eye centers, so the tool follows a tilted head by design, but a large tilt still costs accuracy at the edges of the face. Keep your head roughly level.

Camera distance

Close-up phone shots exaggerate whatever is nearest the lens, so a face shot at arm's length and the same face shot from two meters do not produce the same proportions. Shoot at eye level, arm's length or further.

Expression

A half-smile, a raised eyebrow or a jaw set slightly forward all move the mouth and brow pairs, and they are asymmetry, just not the structural kind you were asking about. Neutral, mouth closed.

What the model can see

Landmarks get placed where there is a visible edge, so side lighting makes the shadowed side read differently. Use even, front-on light, pull hair back from your jaw and brow, and take off glasses that sit across the outer eye corners.

Facing the camera

Turned a little

Same face, same measurement. Turning the head moves the midline toward the far side and compresses that half, so the two readings stop matching — without anything about the face having changed.

This tool refuses to score a photo that is turned or framed too far off, rather than returning a confident number from a bad input. A rejection is more useful than a wrong answer. And if you plan to compare two readings later, write down how you took the first one: reproducing the setup matters more than perfecting it.

What this face symmetry test will not do

It will not tell you where you rank

There is no percentile on this page, because a percentile requires a reference sample and a stated one at that.

It will not diagnose anything

It measures distances between landmark points in a photo. It cannot see nerve function, bite, breathing, or anything under the skin, and a symmetry score is not a screening result.

It will not settle a disagreement between two other tools

If three sites gave you three numbers, all three may have measured correctly against three different definitions.

It will not tell you your face is good or bad

Symmetry is one measurable property of a face. What that property is worth is not a measurement question, and any tool that answers it for you has stopped measuring.

Questions people actually ask

Now do it with your own face

You have read what the number can and cannot tell you. The part worth having takes one photo: which of your five regions is furthest off, and the two mirrored halves that show it.

Upload a photo