The quick version: you can check face symmetry four ways — the mirror test, the mirrored-photo test, manual measurement, and automatic landmark analysis. Only the last two produce a number, and all four are worthless if the photo you start from is taken badly. Get the photo right first; everything else is arithmetic.
First: take a photo that can actually be measured
This step decides your result more than the method you pick. Every technique below measures what the image shows, and a bad image manufactures asymmetry that isn't on your face.
- Camera at eye level, not above or below. Height changes tilt the facial plane and compress one side.
- Face square to the lens. A few degrees of rotation shortens one half in the frame. This is the single most common source of fake asymmetry.
- Distance of 1.5 metres or more, using the rear camera or zoom. Close-range wide lenses stretch central features, and if you're even slightly off-centre that stretch lands unevenly.
- Even, soft light — window light facing you is ideal. A single side light darkens one half and moves where any measurement tool places contour points.
- Neutral expression, hair off the face, no glasses, no filters.
Take two or three like this. If your result is consistent across them, it's real. If it swings, the photos were the variable.
Method 1 — the mirror test
Stand square to a mirror in even light and look at paired features one at a time rather than at your face as a whole: brow heights, eye corners, nostril edges, mouth corners, jaw contour. Comparing pairs is the trick — looking at the whole face, your brain normalizes and you see nothing.
What it tells you: whether you have any visually obvious asymmetry. What it doesn't: anything measurable. And there's a catch — a mirror shows your face reversed. You are habituated to the reversed version, which is why your photos look subtly "wrong" to you and normal to everyone else. The mirror is the least objective instrument you own for this.
Method 2 — the mirrored-photo test
Take your straight-on photo, split it down the midline, and mirror each half onto itself. You get two composite faces: left-left and right-right. The more different those two composites look, the more asymmetry you carry. Any photo editor with a flip tool does this in about a minute.
What it tells you: a vivid, intuitive sense of where your asymmetry sits. What it doesn't: a number, and it exaggerates. Both composites will look strange even on a very symmetric face, because a perfectly mirrored face is subtly uncanny to look at — which is itself worth knowing before you draw conclusions from it.
Method 3 — measuring it yourself
Draw a vertical midline through the nose bridge and the philtrum, then measure the horizontal distance from that line to each of a matched pair: outer eye corners, mouth corners, jaw points. Compare left to right, and divide the difference by your face width to get a proportion rather than a raw millimetre figure — a 2mm gap means far more on a narrow face than a wide one.
What it tells you: a real, if coarse, number. What it doesn't: consistency. Your own marker placement is the input, so two attempts on the same photo give two answers. That's the same weakness that makes manual line-drawing beauty calculators unreliable.
Method 4 — automatic landmark analysis
This is what our face symmetry test does: it detects 50+ facial landmarks automatically, establishes the midline from the nose bridge and philtrum, then measures how far every mirrored pair deviates from its reflected position and normalizes the total against face width.
What it tells you: a reproducible score — the same photo always returns the same number — plus a per-feature breakdown showing which pairs carry the asymmetry, rather than one blended figure. What it doesn't: whether you're attractive. Symmetry correlates with rated attractiveness far more weakly than internet folklore suggests, which is why we score it as one of 14 features rather than the headline.
You'll see this called a symmetry checker, calculator, analyzer or scanner — same operation, different labels. What actually differs between tools is whether landmark placement is automatic, whether you get a breakdown or a single percentage, and whether the tool discloses what population your score is compared against.
Reading the number
Most people land in the 80s. Mid-90s and above is unusually balanced. Below 70 usually traces to one feature pair rather than a globally lopsided face — and more often than not, to the photo. We break the bands down properly in what counts as a good face symmetry score.
Worth holding onto: measurable asymmetry exists in every human face. It's a description of ordinary anatomy, not a defect, and most of it is invisible in conversation. If you're checking because something looks like it changed — suddenly, or progressively, especially alongside drooping, numbness, or trouble moving one side of your face — that's a question for a doctor, not a face-rating tool.
Check yours
The free face symmetry checker runs the landmark analysis in your browser in a few seconds, no signup, and returns the per-feature breakdown alongside 13 other facial categories so you can read the number in context.