Psychology

Reading the Mind in the Eyes Test: What It Measures and How to Read Your Score

By The MentalismPro TeamUpdated

As an Amazon Associate we earn from qualifying purchases. How we choose products.

Black and white close-up of sculpted eyes, the region the Reading the Mind in the Eyes Test asks you to read

The Reading the Mind in the Eyes Test, usually shortened to RMET or the Eyes Test, shows you a series of photographs cropped to a person's eyes and asks which of four words best describes what that person is thinking or feeling. It is one of the best known tests in psychology of how well people read others, and one of the most debated.

This page explains what the test is, where to take the official free version, how scores are described in the research, and what critics say it measures. We do not host the test itself: the Autism Research Centre at the University of Cambridge publishes it for non-commercial research use and does not allow it to be adapted without permission, so the page sends you to the source instead.

What the Eyes Test Is

The test was created by Simon Baron-Cohen and colleagues at Cambridge, and the version most people take is the revised one described in a 2001 paper by Baron-Cohen, Sally Wheelwright, Jacqueline Hill, Yogini Raste and Ian Plumb in the Journal of Child Psychology and Psychiatry. Its aim was to measure how well adults can infer mental states from a small part of the face, an ability often called theory of mind.

How the Items Work

The revised version has 36 items. Each one shows a pair of eyes and four words such as a feeling or an attitude, and the respondent picks the one that fits best. Each item is marked right or wrong against a fixed key, and the total is the score out of 36.

The test is short, free to download and easy to score, so it spread across psychology, psychiatry and neuroscience. It also gave a pleasing way to talk about a skill that is hard to measure. That popularity is part of why its limits matter, which later sections cover.

How to Take It

The Autism Research Centre lists the adult Eyes Test on its own site, with instructions and the test pages. You can find it at the Autism Research Centre's Eyes Test (Adult) page. Use that version rather than a copy from a quiz site, because copies often change the photographs, the word lists or the key.

A Sensible Way to Take It

Read the instructions first and take the test in one sitting, in a quiet place, without a time limit. Pick the word that fits best, even if two seem close, and do not look anything up while you answer. When you finish, score it only against the official key.

What Not to Do

Do not memorize the answers and retake the test, because a second score then measures memory. Do not take it repeatedly hoping to improve a number, and do not treat a copy found online as a source of truth for your score.

How It Is Scored

One point goes to each item where your chosen word matches the key, so the score runs from 0 to 36. The research then compares totals across groups, not across single people.

Typical Scores in the Research

In the 2001 study, normal adult groups scored roughly 26.2 to 30.9 out of 36, which is 74% to 86% correct, and adults with Asperger syndrome or high-functioning autism averaged 21.9, about 61%, as reported in a 2015 analysis by Sally Olderbak and colleagues in Frontiers in Psychology. In that same analysis, an online adult sample averaged 27.2 out of 36 with a standard deviation of 3.8.

What Above and Below Those Figures Mean

A score near the middle of those group averages is ordinary. A score a few points below or above says little by itself, because group averages cannot tell you what one person's number means. The 2001 study compared groups, which is different from diagnosing an individual, and the test is not a diagnostic tool on its own. If you have questions about your own results, a qualified clinician is the right person to ask.

What the Test Does and Doesn't Measure

The Eyes Test has been used for more than twenty years, and a number of studies have asked what it actually measures. The answers are more mixed than its reputation.

The Vocabulary Problem

Olderbak and colleagues found in 2015 that the full 36-item test is not homogeneous and has poor internal consistency, and that it correlates strongly with vocabulary, at 0.54 in their first study. That link means word knowledge plays a part in getting items right. They proposed a 10-item short form with better consistency and said future versions should depend less on vocabulary.

Recent Doubts About the Structure

In 2025, Wendy Higgins, Victoria Savalei, Vincent Polito and Robert Ross published an analysis in Assessment of nine large non-clinical datasets, with from 558 to 9,267 participants. A single-factor model did not fit well in any of them, some items failed the original validation criterion in several datasets, and apart from a handful of easy items, participants sometimes disagreed substantially about which answer was correct. They concluded the test may not be a valid measure of social cognitive ability. They note that their samples did not include clinical or autistic participants, so their findings apply to the general population.

What It Can Still Do

None of this makes the test useless. It remains a short, free way to look at how people choose between mental-state words for photographed eyes, and it is a useful example of how a famous test can be questioned. What the evidence does not support is reading it as a clean score for empathy or for how well you read people in real life.

Reading Eyes in Mentalism

Mentalists and magicians have long claimed that the eyes give away thoughts, and they use that belief on stage. The honest version of the claim is smaller than the stage one. People do show signals in their faces, but a single pair of cropped eyes, like those in the Eyes Test, carries far less than a person talking to you in context.

What Performers Actually Use

A performer reads the whole situation: how the spectator stands, how fast they answer, where they look when they are asked a question, and how they react to a guess. A single photograph of eyes tells almost nothing about any of that. Our guides to micro expressions and how to read people cover what face and body cues are worth and where the myths begin.

Eyes and Lying

The belief that eyes reveal a lie is one of the most common. What works and what does not is in how to tell if someone is lying, and the way readers use guesses and reactions to seem insightful is in cold reading. If the question of fast versus careful judgment interests you, the intuition test is a short companion to this page.

A Book to Read Next

In the 2001 study, ordinary adult groups averaged roughly 26 to 31 out of 36, so a score in the high twenties is typical. A 2015 analysis of an online adult sample found an average of 27.2. There is no pass mark, and single scores say little because the test was built for group comparisons.
No, a score cannot diagnose anything. The 2001 study compared group averages, with adults with Asperger syndrome or high-functioning autism averaging about 22 out of 36, but groups overlap and an individual score does not identify anyone. Diagnosis needs a clinical assessment, and the test's own validity is debated.
Looking at more faces with feedback may build familiarity, but practicing the Eyes Test itself mostly trains you on its words and photos, and a repeat score reflects memory. Real-world skill comes from watching people in context, such as voice, timing and what was said, which a cropped photo cannot supply.

Where to go next: micro expressions for what faces can and cannot show, how to read people for body language in context, and the intuition test for a look at quick judgments.

Free newsletter

Mentalism, made modern. In your inbox.

Occasional emails from the MentalismPro team: a technique to practice, a routine idea, and the new guides and books worth your time. No spam, unsubscribe any time.