Face shape statistics.
The numbers, in context.

How common is each face shape?

Oval accounts for almost half of this sample. Oblong and square are close behind one another, at 14.5% and 14.3%. Diamond appears least often among the six available labels. Every percentage below uses the same denominator: 2,048,842 analyses from August 1 to October 31, 2026. Source: The Face Report [1].

Why oval leads, and what “rarest” means

The source describes oval as the middle profile in its classifier: faces without strongly long, broad, angular or tapered proportions tend toward that label. A large oval share therefore reflects the classification rules as well as the photos people submit.

Diamond’s 5.4% is a result within this six-label system. Triangle and pear are absent, so neither can appear at the bottom of its ranking. A different set of labels can change which shape seems most common or rarest. Rarity also says nothing about attractiveness.

Does the distribution stay stable by month?

The broad ordering holds across the three months: oval leads, and heart and diamond remain the two smallest groups. Second place changes: square is ahead of oblong in August and September, while oblong moves ahead in October.

Oval rises from 48.1% to 49.7%, a change of 1.6 percentage points. This is a descriptive shift in submitted scans. It does not show that people’s underlying facial anatomy changed over three months. The source reports one classifier version throughout the period; repeat scans and changes in who used the tools still affect the comparison.

How common are high symmetry scores?

The same source publishes 2,015,960 analyses with a stored symmetry band. About 9.8% score at least 90 on its 100-point scale. The largest band is 80–89 at 32.3%. These scores belong to that tool’s scoring system; they are not interchangeable with scores from another symmetry test. Source [1].

The 95–100 band is 0.4% of the retained counts, with an important limitation: the source only stores day/tool/band groups containing at least 10 analyses. Smaller groups are omitted, which particularly undercounts the highest band. The symmetry denominator is therefore smaller than the face-shape denominator. Head pose and photo quality can also influence a score.

What academic research measures instead

Anthropometric studies often use the facial index: measured facial height divided by cheekbone width, multiplied by 100. Height typically runs from the nasion to the gnathion. Their categories describe relative breadth and length; they cannot be translated directly into oval, heart or diamond labels.

Pandeya and Atreya studied 155 medical and nursing students in Nepal, using calipers. Mesoprosopic was the largest category at 38.06%. The table preserves the published counts and percentages. Read the study [2].

A separate study by Jeremić and colleagues measured 700 adults aged 18–65 in central Serbia. Its reported mean facial index was 93.68, compared with 87.44 in the Nepal student cohort. Differences in recruitment, measurement and participants matter; these two cohorts cannot establish a national or global ranking. Read the study [3].

Facial index values are reported mean ± SD. Study [2] prints an inverted formula in its methods, while its index values follow height/width; we retain its published results and flag the inconsistency in the downloadable notes. Neither study supplies a worldwide distribution.

Five checks before quoting a percentage

1. 01

   Check the unit.

   The main table counts analyses. The source reports that at least 44.5% followed another scan in the same session, tool and day. A repeat scan is another observation.

2. 02

   Keep the denominator attached.

   Face shapes use 2,048,842 analyses. Symmetry uses 2,015,960 retained results. Research rows count participants. These denominators answer different questions.

3. 03

   Name the classification system.

   The Face Report uses six profiles and a MediaPipe landmark model, comparing height/width, forehead/cheekbones, cheekbones/jaw and jaw angle. A seventh label or different thresholds can change the distribution.

4. 04

   Consider the sample and the photo.

   Free-tool users choose to submit a photo. Head rotation, lens distance and obscured contours can change the measurements. Large samples still carry selection and measurement bias.

5. 05

   Respect missing information.

   The source does not store age, sex or country with a free analysis. These counts cannot support demographic breakdowns. The research cohorts supply context, not a way to fill those gaps.

The source period spans 92 complete UTC days; counts were extracted on November 12, 2026. We checked and compiled the sources on November 13. The two original CSVs are mirrored unchanged. Monthly shares are calculated from their counts, and research facts are transcribed into separate tables. No datasets are pooled.

Our own report is still ahead

Download the data behind the tables

Choose a CSV for a spreadsheet, or download the combined JSON for every table and its provenance. The files contain aggregate statistics. Original photos, landmarks and individual user records are not included.

Snapshot: November 13, 2026. Download the source notes alongside the tables when citing or reusing the numbers. Percentages may differ slightly from 100% when rounded.

Sources & attribution

1. [1] The Face Report (2026)

   Face shape statistics: 2,048,842 analyses, August 1–October 31, 2026. Face-shape counts and symmetry bands © The Face Report, reused under CC BY 4.0. Retrieved November 13, 2026.

   Original aggregate CSVs are unchanged; this article’s charts, monthly calculations and commentary are by PSL Scale. Full original source identifiers are retained in the downloadable attribution notes. Reuse does not imply endorsement.

2. [2] Pandeya A; Atreya A (2018). Variations in the Facial Dimensions and Face Types among the Students of A Medical College.

   DOI: 10.31729/jnma.3470. 155 medical and nursing students (72 male, 83 female). Numerical findings compiled from tables 1–3; no individual records are reproduced.

3. [3] Jeremic D et al. (2013). Anthropometric study of the facial index in the population of central Serbia.

   DOI: 10.2298/ABS1303163J. 700 adults (360 male, 340 female), ages 18–65. Numerical findings compiled from abstract; no individual records are reproduced.

Frequently asked questions