Free IQ Test

How a thirty-item IQ test is built, and what your number actually means

An intelligence test is not a quiz about facts. Nothing on this page rewards you for knowing the capital of a country, the date of a war or a word you happened to meet in a novel. Every item instead presents a small, self-contained problem whose rule is visible in the item itself, and asks whether you can find that rule quickly enough to use it. That design choice is the whole discipline in miniature: a good test measures the machinery, not what the machinery has been fed. Thirty items across five families, with a twenty-five minute ceiling, is the shortest format that reliably separates people rather than merely amusing them.

Test length buys precision, and it buys it at a decreasing rate. Ten items produce a number so noisy that two people of identical ability routinely land fifteen points apart. Thirty items cut that error roughly in half. Sixty would halve it again, but almost nobody finishes a sixty-item test taken voluntarily on a website, and an abandoned test measures nothing at all. Thirty sits at the point where the curve of precision and the curve of patience cross.

The five families exist for a different reason. Reasoning ability is not one thing, and any single format quietly measures something extra alongside it. Pure number tests reward arithmetic fluency. Pure verbal tests reward vocabulary and, unavoidably, the language you grew up speaking. Pure spatial tests reward practice with maps, drawings and video games. Mixing matrices, number series, verbal analogies, odd-one-out items and rotation figures means those side effects partly cancel, so what survives in the total is closer to the shared factor that runs through all of them. Professional batteries follow the same logic when they build a full-scale score out of several subtests rather than trusting any one of them.

What each family is really asking

Matrix puzzles — ten of the thirty — show a three-by-three grid whose cells vary by shape, count, shading or rotation, and ask which tile completes the missing corner. They come closest to measuring reasoning stripped of language and culture, which is why they carry the largest share here. Number series give a run of values that follows one hidden rule; the difficulty is almost never the arithmetic but deciding which family of rule you are looking at before you start calculating. Verbal analogies test relational precision: you must name the relationship inside the first pair exactly, then find the pair that shares it, while resisting options that are merely related to the same topic. Odd one out items are the inverse — you must invent several candidate rules, then find the one under which exactly one member fails. Spatial rotation figures ask whether a shape can be turned into another without lifting it off the page; the trap is always a mirror image, which looks right until you try to rotate it.

Reading your score

The raw total is mapped onto the standard scale where 100 is the population average and one standard deviation is fifteen points. Because thirty items still leave real measurement error, the honest reading of your result is a band of roughly eight points on either side of the number you see.

Getting an honest number out of yourself

• Take it once, rested, in a single sitting. Every later attempt measures memory as much as reasoning.
• Answer every item. There is no penalty for a wrong answer, so a blank simply throws away a chance.
• Budget about fifty seconds an item. If one resists you past a minute, guess, move on, and come back with the Back button if the clock allows.
• Use a laptop or tablet if you can. Matrix and rotation figures are small on a phone, and squinting costs seconds you will want later.
• Look nothing up. The moment you search for a series rule you are measuring your search skills.
• Do not stop the clock. A paused test produces a number that is not comparable to anyone else's.

How this test differs from the shorter ones

Think of the tests on this site as widening lenses on the same object. The quick five-minute test tells you which broad band you probably fall in and little more; it rewards decision speed heavily, and careful people are undersold by it. The culture fair test drops language entirely and spends every item on visual pattern reasoning, which makes it the fairest option if English is not your first language. This test sits between them in spirit and above both in precision: it keeps the verbal and numerical content that a full battery would include, but balances it so that no single weakness decides the total. If you intend to take more than one, take this one first — it is the one most damaged by having already seen items of a similar shape.

The limits worth stating plainly

No test taken alone in a browser is a clinical assessment. A psychologist administering a clinical or a clinical battery controls the room, watches how you approach each problem rather than only whether you solved it, and compares you against a norm sample stratified by age and background instead of against whoever happened to visit a website. Motivation, sleep, caffeine, screen size and a single interruption can each move an online result by several points. Treat your score here as a good estimate for curiosity and practice, as a useful signal when two formats disagree sharply, and as no basis whatsoever for a diagnosis, a school placement or a hiring decision. Any site that claims otherwise is selling something.