Here is exactly how IQTests turns a set of answers into an IQ score. We publish the method in full because a number you can’t interrogate isn’t worth much — and because transparency is the honest alternative to a black box.
The questions
The test is built from reasoning items rather than trivia. Knowing a capital city measures what you have learned; these items measure how you work out something new. The core is non-verbal, matrix-style pattern problems in the tradition of Raven’s Progressive Matrices, chosen because shape-and-sequence puzzles lean far less on vocabulary or first language than a word quiz would, so the result travels better across backgrounds. Alongside the matrices, the result reads across four reasoning areas — abstract, verbal, numerical and spatial — so a single number never hides where your strengths actually sit.
From answers to a score
Your raw score is simply how many items you get right. That raw count is then placed on the deviation-IQ scale, the same framing modern tests use: the average is set to 100 and a standard deviation is 15 points. In plain terms, a result near 100 sits in the middle of the reference distribution; move 15 points in either direction and you have crossed roughly one standard deviation. From that position on the normal curve we derive your percentile — the share of people your result lands above.
How far to trust the number
For most people, a careful online reasoning test lands close to where a formal test would put them, and it correlates with gold-standard instruments well enough to be genuinely useful. But it is an estimate with a margin of error, not a point-perfect measurement. Treat your score as a band a few points wide rather than a single exact figure, and treat it as one honest data point about how you handled these puzzles — not a verdict on who you are.
What can move a score
Scores are not fixed to the decimal. Fatigue, a noisy room, a tiny phone screen, or answering at 1 a.m. can each cost you several points. Practice with this style of puzzle nudges results upward. And across whole populations, average performance drifted upward for generations — the Flynn effect — which is why deviation scales are periodically re-centred back to 100. Take the test rested and undistracted if you want your cleanest read.
Review and updates
This method is reviewed by Jennifer Huber, Ph.D., a specialist in cognitive assessment and psychometrics, and revisited as the underlying research changes. When we adjust scoring or wording, the change is reflected here.
An estimate, not a diagnosis. IQTests gives an educational estimate of reasoning ability. It is not a clinical or psychological assessment, and it should not be used for diagnosis, placement, or any high-stakes decision. Our terms of use cover this in full.
For the story behind the project and the team, see about IQTests. The clearest way to see the method in action is to take the test and read your own breakdown.
Reviewed by Jennifer Huber, Ph.D. · Last updated July 2026