What Is My Face Shape? Find Yours With 4 Measurements (No Photo Needed)

What is my face shape? The two-minute answer

Your face shape is set by four numbers, forehead width, cheekbone width, jaw width, and face length, and by how they compare to each other. The fastest way to find yours is our face shape quiz: about two minutes, a handful of questions about what you can see in a mirror, and no photo. Nothing you enter leaves your browser.

The quiz works because you do not need millimeter precision to land in the right category. You need to know three things: whether your face is clearly longer than it is wide, which of the three widths is largest, and whether your jaw is angular or curved. Those are questions you can answer by looking, and the quiz walks you through them in the right order.

If you want the precise version, the kind you can write down in centimeters and reuse for glasses sizing, hairstyle guides, or a salon consultation, the tape-measure method below takes about ten minutes and gives you numbers instead of impressions. Most people do both: quiz first for the quick answer, tape second to confirm it.

Either way, the result is a starting point, not a verdict. Face shapes are descriptive categories, and what you do with yours should follow your actual features. More on that in the section on blended shapes.

Why do most people guess their face shape wrong?

Because the two tools everyone reaches for first, the mirror and the front camera, both distort the answer, and because hair hides the landmarks that define it. Guessing from a glance gets the compact shapes confused with the long ones and the forehead confused with the hairline.

Start with the mirror. You look at a mirror the way you look at a person: eyes first, center of the face, from a comfortable distance. You almost never study your own jaw width or the true proportion of length to width, and the slight downward or upward tilt most people hold flattens or stretches the apparent length of the face. The result is that people reliably judge their face as longer or shorter than the tape says it is.

The front camera is worse, in a measurable way. A face photographed at typical selfie distance, around 30 centimeters, is geometrically distorted: features closest to the lens are enlarged relative to the rest of the face. A published model of this effect found that a selfie taken at 12 inches makes the nasal base appear about 30 percent wider than a photo taken at standard portrait distance research-backed [4]. The same geometry compresses the apparent width of your jaw and the outer face, which is exactly the region that separates a round face from an oval one or a square from a rectangle.

Then there is the hairline problem. Hair worn over the temples makes the forehead read narrower than the bone underneath; volume at the sides makes the midface read wider; bangs erase the top of the face entirely, so people substitute the eyebrow line for the hairline and lose two or three centimeters of length. What you are categorizing when you glance in the mirror is your current haircut’s silhouette, not your face.

Finally, the reference lists people compare themselves against do not agree with each other. Some systems use six shapes, some seven, some eight or nine, and the same celebrity appears under different labels on different sites. We use a seven-shape system, laid out in the 7 face shapes, and we are explicit that the taxonomy itself is a styling convention rather than an anatomical fact convention. The measurements below are the stable part; the labels are bins drawn on top of them.

How do I measure my face shape? The four measurements

You need a soft tape measure, a mirror at eye level, and your hair pulled completely off your face. Measure four things, forehead width, cheekbone width, jaw width, and face length, write each number down in centimeters, and the next section turns them into a shape.

A few setup notes before the first measurement, because they prevent most of the errors:

• Pull all hair back, including bangs, side pieces, and anything covering the ears. You need to see your hairline and feel your jaw corners.
• Face the mirror square-on with your chin level. A tilted head changes the apparent endpoints.
• Keep your face relaxed. Clenching, smiling, or raising your eyebrows moves soft tissue over the landmarks. The one exception is noted in step 3.
• Hold the tape taut and straight between the two endpoints for the width measurements. You are measuring straight-line distances, not tracing the curve of your face. Letting the tape hug the contour inflates every width and is the single most common source of nonsense numbers.
• No soft tape? Use a piece of string: hold it between the landmarks, pinch where it ends, and lay it against a ruler.

The landmarks below come from the standard vocabulary of facial anthropometry, the measurement system compiled by Leslie Farkas, whose reference work defined the linear distances and ratios still used in craniofacial research research-backed [1]. You will be finding consumer-friendly versions of four of his landmark pairs. You do not need the Latin names to measure well, but they are included so you can see this is not made up.

Step 1: Forehead width

What to measure: the widest horizontal span of your forehead, straight across from one side to the other.

Where the landmarks are: the widest part of the forehead usually sits roughly halfway between your eyebrows and your hairline. Your endpoints are where the flat plane of the forehead ends and curves back toward the hairline on each side, just above the tail of each eyebrow. In anthropometric terms you are approximating the span between the temporal crests, near the landmark pair Farkas called frontotemporale [1]. Practically: place a fingertip above the outer end of each eyebrow and slide it outward and slightly up until you feel the surface start to turn away from the mirror. Measure straight between those two points.

Common error: measuring at the hairline itself. Hairlines curve inward at the top corners, so a measurement taken there returns the width of your haircut’s opening, not your forehead. Stay at the mid-forehead level where the bone is widest. Second error, specific to this step: pressing the tape into wisps of hair at the temples; tuck them back first.

Step 2: Cheekbone width

What to measure: the widest span of your face across the cheekbones, straight across, passing over the bridge of your nose.

Where the landmarks are: you are looking for the most outward-jutting point of each cheekbone, the landmark called zygion; the distance between the two is the bizygomatic width, the standard “face width” of the anthropometric literature research-backed [2]. Find it by feel: place your index fingers just below the outer corner of each eye and press gently. You will feel a bony ridge. Follow it outward, toward the top of each ear, until you find the point that sticks out farthest to the side, usually two to three centimeters forward of the ear canal. Hold the tape straight from point to point.

Common error: measuring the cheeks instead of the cheekbones. The soft fullness of the cheek sits lower, and on most faces it is not the widest bony point. If you measure with a relaxed face and stay on the ridge you found by feel, you are on bone. Smiling is the other classic mistake: it lifts and widens the soft tissue and can add a centimeter that is not structural.

Step 3: Jaw width

What to measure: the straight-line distance between the two corners of your jaw, the angle below and slightly forward of each earlobe.

Where the landmarks are: this is the bigonial width, between the left and right gonion, the angle of the mandible [1]. This is the one step where you should briefly clench your back teeth: the masseter muscle bulges and the corner of the jaw becomes easy to find under your fingers. Locate the angle on each side, unclench, and measure straight across from corner to corner, with the tape passing in front of your chin through the air, held taut.

Common error: wrapping the tape along the jawline from ear to chin to ear. That traces the length of the jaw’s curve, a different and much larger number, and it will wreck the comparison in the next section. You want the straight-line width, as if you were measuring the face in a photograph. If your jaw number comes out close to your cheekbone number times 1.5, you almost certainly wrapped.

Step 4: Face length

What to measure: the vertical distance from the center of your hairline to the bottom of your chin.

Where the landmarks are: the top point is the middle of your hairline where it meets your forehead (trichion, in the anthropometric vocabulary); the bottom point is the lowest point of your chin (menton) [1]. Look straight into the mirror, chin level, and run the tape down the midline of your face, over the nose, keeping it straight rather than pressed into the curve of the nose bridge.

Common error: starting at the wrong top point. A widow’s peak dips below the true hairline, so start at the line itself on either side of the peak, not at its tip. If your hairline has receded or is not obvious, use the point where the vertical plane of your forehead starts to curve back toward your scalp; as a fallback, the forehead on most adults occupies roughly the top third of the face, so you can sanity-check your number against that. What matters most is picking a point and using the same convention if you ever remeasure.

As a sanity check on all four numbers: for most adults, cheekbone width lands somewhere between roughly 12 and 15 centimeters, and face length between roughly 17 and 20 [2]. If one of your numbers is wildly outside that neighborhood, the tape probably wandered, and the usual culprits are the wrap-around jaw error or a hairline starting point at the crown.

If you want these four steps as a printable one-pager, landmark diagrams plus write-in boxes for your numbers, it is here: the measuring guide. Print it, measure once, and you will never need to do this from memory again.

How do I turn four numbers into a face shape?

Answer three questions in order: is your face clearly longer than it is wide, which of the three widths is largest, and is your jaw angular or curved? Those three answers map to one of the seven standard shapes, with rectangle appearing as the angular variant of oblong, and the table below compresses the whole decision into one lookup.

Here is the logic in prose, in the order that resolves the most ambiguity soonest.

Question 1: length versus width. Divide your face length by your cheekbone width. If the ratio is around 1.5 or higher, you are in the long family: oval, oblong, or rectangle. If it is closer to 1.0 to 1.3, meaning your face is nearly as wide as it is long, you are in the compact family: round or square. Heart, diamond, and triangle can occur at either length, because they are defined by the width pattern rather than the ratio.

Question 2: the widest point. Rank your three widths. If your forehead is the widest of the three, you are looking at a heart shape. If your jaw is the widest, that is a triangle (sometimes called pear). If your cheekbones are the widest by a clear margin, with both the forehead and the jaw noticeably narrower, that is the signature of a diamond face shape. If all three widths land within about a centimeter of each other, your face reads as a column: square or rectangle depending on length. Cheekbones slightly widest, with forehead a little wider than jaw, is the classic oval and round pattern.

Question 3: the jaw and chin. This is the tiebreaker. An angular jaw, one where you can point to the corner you found in step 3 just by looking, pushes the answer toward square and rectangle. A curved jaw that flows into the chin without a visible corner pushes toward round and oval. A chin that narrows to a point is the marker for heart and diamond; a broad, squared-off chin belongs to square and triangle faces.

The same logic as a lookup table:

Your measurement pattern	Widest point	Jaw and chin	Likely shape
Length about 1.5x cheekbone width, forehead slightly wider than jaw	Cheekbones	Gently curved, rounded chin	Oval
Length and cheekbone width nearly equal	Cheekbones	Soft, curved, no visible corner	Round
Length and width nearly equal, all three widths within about 1 cm	Even column	Angular corner, broad chin	Square
Length clearly greatest, all three widths within about 1 cm	Even column	Angular corner	Rectangle (angular oblong)
Length clearly greatest, widths even, jaw softly curved	Even column	Curved, softer chin	Oblong
Forehead is the widest of the three	Forehead	Narrowing to a pointed chin	Heart
Cheekbones widest by a clear margin, forehead and jaw both narrow	Cheekbones	Pointed chin	Diamond
Jaw is the widest of the three	Jaw	Broad, strong lower face	Triangle

Two notes on reading the table honestly. First, the thresholds are conventions, not physical constants: the 1.5 ratio, the one-centimeter tolerance, and even the number of named shapes vary between systems, which is why different guides can file the same face differently convention. We keep the full definitions, with the comparisons between neighboring shapes, on the 7 face shapes page. Second, the underlying measurements are the trustworthy part. The anthropometric literature treats them as continuous distances and indices with population norms [2], and the named shapes are a styling shorthand layered on top. If your numbers are clear but the label feels ambiguous, trust the numbers.

What if I’m between two face shapes?

Then you are typical. Faces are continuous, categories are bins, and a large share of faces sit near a boundary. Pick the shape whose defining feature is strongest on your face, treat it as your primary label, and read the neighboring shape’s guide as well.

The boundaries people land on most often are predictable from the decision logic. Oval and oblong differ mainly in the length ratio, so a face at 1.45 is genuinely both. Round and square share the compact proportions and differ only in the jaw corner. Heart and diamond share the pointed chin and differ only in whether the forehead or the cheekbones win the width ranking; if those two numbers are within a few millimeters on your face, the label is a coin flip and nothing about your features changes with the coin.

The practical resolution is to let styling logic follow features rather than labels, which is how most competent stylists already work convention. Guidance written for square faces is really guidance for an angular jaw; if you have an angular jaw on an otherwise long face, that guidance applies to you regardless of whether your quiz result said rectangle. Guidance for diamond faces is mostly about the width contrast between cheekbones and the zones above and below them; a heart-diamond blend with prominent cheekbones should read it. The label is an index into advice, not a diagnosis, and blends simply mean you get to index into two chapters instead of one.

When you take the quiz, borderline patterns are reported as exactly that: a primary shape plus the neighbor you share features with. We would rather tell you “oval with oblong length” than force a clean answer the tape does not support.

One more reason not to agonize over a boundary call: the inputs themselves drift a little. Soft tissue over the jaw and cheekbones changes with weight, and it redistributes gradually with age, so a face can migrate between adjacent bins over the years while the skull underneath stays put. A label you were given at twenty is worth rechecking; the four measurements take ten minutes.

How accurate are face shape photo apps?

Upload-your-photo detectors are convenient, but their results disagree with each other often enough that you should treat any single app’s answer as a guess, not a measurement. The evidence on photo-based facial measurement is genuinely mixed evidence, and the disagreement has three specific causes.

First, 2D photos are weakest exactly where face shape is decided. A validation study comparing measurements taken from standardized 2D photographs against direct measurement of the same 96 faces found that seven of ten facial dimensions were accurate, but the two with both statistically and clinically significant errors were bizygomatic breadth and bigonial breadth research-backed [3]. Those are the cheekbone width and jaw width from steps 2 and 3, two of the three numbers the whole classification runs on. And that study used controlled, standardized photography; a phone snapshot with uncontrolled distance, tilt, and lens adds error on top.

Second, the camera geometry problem from earlier applies in full. Selfie-distance photos enlarge central features and compress the face’s outline relative to a portrait taken at proper distance [4]. Two photos of the same face, one at arm’s length and one propped on a shelf two meters away, can present measurably different width ratios to whatever algorithm receives them. Apps rarely control for this, and users rarely know it matters.

Third, the algorithms themselves are trained on different answers. Published face shape classifiers report accuracies from roughly 70 percent for classical landmark-plus-SVM approaches to about 85 to 90 percent for CNN and transfer-learning models, each measured against its own labeled dataset [6]. Those datasets use different shape taxonomies, five categories in some, seven or more in others, and the labels are assigned by humans who disagree at the same boundaries you might. The same survey literature notes that oval, the most common label, is also the most frequently misclassified, and that misclassification rates differ across ethnic groups underrepresented in training data [6]. Two apps built on different datasets are not making errors relative to a shared truth; they are answering slightly different questions.

This is why our tools work the way they do. The quiz asks you about direct observations, which point of your face is widest, whether you can see a jaw corner, because you have information about your own face that a single compressed photo does not carry. The tape method gives you the ground-truth numbers that photo pipelines approximate. And nothing about either requires uploading your face to anyone’s server: the quiz runs entirely in your browser, no photo, no account, and your answers never leave your device. If you have already measured, you can enter your four numbers directly and skip the questions.

None of this means photo tools are useless. A well-taken photo, straight-on, distant, hair back, is a reasonable cross-check, and the research on photogrammetry shows most facial dimensions survive the trip through a lens [3]. It means a photo app should be the second opinion, not the primary source, and that when an app disagrees with your tape measure, believe the tape.

What should I do with my face shape result?

Write down the shape and the four numbers, then go to the guide for your shape: the label is only useful for what it unlocks. Your numbers are reusable for glasses sizing, hairstyle consultations, and any future remeasure.

Start with the 7 face shapes if you want to see your result in context, how your shape sits next to its neighbors and what specifically separates them. If your result was diamond, or a blend involving prominent cheekbones, the diamond face shape guide covers the width-contrast logic that drives most styling advice for that pattern. The full set of per-shape guides, comparisons, and styling references lives in the face shapes hub, and each per-shape page links to its adjacent shapes so a borderline result can read both of its chapters side by side.

Keep the numbers somewhere you will find them again. Centimeter measurements of forehead, cheekbones, jaw, and length are exactly what an optician means by frame width fit, and exactly what a stylist is estimating by eye during a consultation. The printable measuring guide has a box for them, which is the low-tech version of what our quiz does automatically: your result is saved in your own browser, on your device, so returning visitors see their shape without re-entering anything, and without any of it ever being sent to us.

One last calibration. Everything downstream of the measurement, which hairstyles, frames, or necklines are conventionally paired with which shape, is convention informed by proportion, not a rulebook. The measurements are real; the anthropometry behind the landmarks is a documented scientific literature [1][2]; sex differences in facial width are measurable and well replicated [5]. What you do with the numbers is a set of choices. Descriptions of what a style does to the appearance of width or length are testable in a mirror; claims about what your face shape means you must or must not wear are not. Use the label as a map, and let your own features cast the deciding vote.