What is the Feline Grimace Scale?

The five features

Each one is scored 0 (absent), 1 (moderate, or you genuinely cannot tell) or 2 (obvious). A feature that cannot be seen clearly is not guessed at — it is left unscored.

Ear positionobservers agree 87% of the time

A comfortable cat faces its ears forward. As discomfort rises the tips pull apart and rotate outwards, until the ears sit flattened and turned away from each other.

Orbital tighteningobservers agree 86% of the time

The eye narrows. A relaxed cat holds its eyes open; a cat in pain squints, and the study draws the line at a gap between the eyelids smaller than half the width of the eye.

Muzzle tensionobservers agree 63% of the time

A relaxed muzzle is round — roughly as tall as it is wide, with full whisker pads either side of the nose. Under tension it flattens and stretches sideways towards an oval. Long or fluffy fur can disguise this.

Whisker positionobservers agree 55% of the time

Relaxed whiskers hang in a loose downward curve. As pain increases they straighten and push forward, away from the face, as though standing on end.

Head positionobservers agree 90% of the time

Judged against the shoulders. Held above the shoulder line is relaxed; level with it is intermediate; dropped below the shoulders, or tilted down with the chin towards the chest, is the strongest sign.

Those agreement figures come from the validation study, and they are worth reading. Muzzle and whiskers are measurably harder to call than ears, eyes and head position — which is why PurrSight gives them less visual weight rather than pretending every feature is equally certain.

The diagrams above are our own illustrations, drawn to show one feature changing at a time. They are not the official reference images — those are © Université de Montréal and are available from felinegrimacescale.com.

How the score works

The scores are added up and divided by the maximum possible, giving a number between 0 and 1. Any feature that could not be scored is left out of both sides of that sum, so a partly hidden face still produces a fair result rather than an artificially low one.

Why it works for cat owners

This is the part that makes the scale unusual. A follow-up study compared how consistently different groups applied it — cat owners, veterinarians, veterinary students and nurses — and found they agreed well with one another. It was built to be usable by people without clinical training, which is exactly why a tool like this one can reasonably be put in front of you.

How PurrSight uses it, and where we differ

PurrSight asks an AI model to score those same five features from your photo, several times over, and combines the runs. It is an emulation of the published method, not the method itself, and the differences are worth knowing:

• The scale was validated with trained human observers looking at a cat in front of them. We use a model looking at one photograph.
• It was validated for acute — sudden, short-term — pain. A cat with a long-standing problem may score low and still be suffering.
• Our own testing shows muzzle tension is where we are weakest: the model under-reads it against the scale's own labelled reference photographs. We would rather publish that than hide it.