How the test works

There are 30 questions in five parts: matrix patterns, odd one out, domino sequences, number series and spatial puzzles. Every question is a picture or a number series, so the test works the same in every language.

Every question was written for this test. Each one is generated from its rule and checked automatically, so it has exactly one correct answer. Within each part, questions get harder as you go.

Each correct answer earns a point; the last two questions in each part, the hardest, earn two. There are 40 points in total.

Your share of the points is converted to the IQ scale, where 100 is the population average and 15 points is one standard deviation. With five or six options per question, guessing alone earns about a fifth of the points, and the conversion places that well below average.

We show a likely range of six points either side of your score, because any single short test carries that much uncertainty.

The conversion is provisional. It has not yet been normed against a large, representative sample, and we will not pretend otherwise. As people take the test, we use their anonymous answers to recalibrate how hard each question really is and where scores fall.

This is an online estimate of reasoning ability, not a clinical or diagnostic assessment. A full clinical IQ test is given one to one by a psychologist and covers more abilities, including verbal ones.

We store your answers, your time and your score under a random link. After the test we ask for your email to send you the result; you only get offers if you tick the box for them. If you buy the report, we store the name you choose for the certificate. Payment is handled by Stripe.

Your score goes on the leaderboard anonymously unless you add a nickname. To improve the test we keep anonymous totals, such as how often each question is answered correctly, never tied to you.