Free · No sign-up · Instant result

Free IQ test

Thirty questions covering figural, numeric, verbal and spatial reasoning. You get a score, the margin of error around it, a breakdown by domain, and a written explanation of every question you got wrong.

  • 30 questions, about 30 minutes
  • No account, no email address
  • Answers never leave your browser
  • Scoring method published in full

What the test measures

The four question types are the ones that load most heavily on general reasoning ability in the published literature, and they map onto four broad abilities in the Cattell–Horn–Carroll model of cognitive ability. Each is scored separately as well as together.

Gf - induction (I)

Figural reasoning

Inferring the rules that govern an abstract pattern, then applying them.

10 questions

Gf - quantitative reasoning (RQ)

Numeric & symbolic series

Detecting the generating rule behind a sequence of numbers or letters.

7 questions

Gc / Gf - language development and deduction

Verbal & logical reasoning

Relational analogies and deductive inference expressed in words.

7 questions

Gv - visualisation (Vz)

Spatial rotation

Mentally rotating a figure and distinguishing rotation from reflection.

6 questions

Why these four, and how each item was built →

A question, worked through

Every figural item is generated from an explicit set of rules rather than drawn by hand, which is what lets each one be checked by machine for a single unambiguous answer. Here is an easy one with the answer shown.

Which figure completes the matrix?

?
A
B
C
D
E
F

Two rules are running at once

Each of the three shapes appears exactly once in every row and every column. Separately, shading is constant along each row and changes between rows. The missing cell has to satisfy both at the same time.

This example is not one of the 30 scored questions.

How the score is worked out

Counting correct answers and multiplying by a constant would treat an easy question and a hard question as equal evidence. They are not, so this test does something more careful.

1

Every question has a difficulty

Each item carries a difficulty and a discrimination value, set from how structurally complex it is - how many rules govern a matrix, how deep the generating rule of a series runs, what kind of inference a verbal item demands.

2

Your answers are fitted to a model

A three-parameter logistic model asks which ability level makes your particular pattern of right and wrong answers most probable, allowing for the fact that a lucky guess on a hard item is always possible.

3

The result comes with error bars

The model returns a range, not just a number. Near the middle of the scale that range is about ±9 points at 95% confidence. Any test that reports a bare number without one is overstating what it knows.

The full method, including its limitations →

What a score like this does and does not tell you

An IQ score is a rank, not a measurement of a substance. It says where your performance on these particular questions, on this particular day sits relative to a reference distribution. That is a genuinely useful thing to know, and it is a much narrower thing than "how intelligent you are".

Scores of this kind predict school and job performance moderately well on average across large groups, and predict very little about any single person. They say nothing about creativity, judgement, diligence, or whether someone is good company. They move with sleep, illness, stress, practice and motivation. And the number itself is anchored to a norm group that shifts across generations.

This is not a clinical assessment

A score from this site cannot diagnose anything, cannot support an application for accommodations, and is not equivalent to a supervised test administered by a qualified psychologist. If you need a score that carries weight somewhere, that is the route.

The limitations are listed in full on the methodology page →

Common questions

Is this IQ test really free?

Yes. All 30 questions, the full score report and the per-question explanations are free, with no account, no email address and no payment at any point. The site is funded by a small number of display ads placed outside the test itself.

How long does the test take?

About 30 minutes. There is a 30-minute limit for the whole test, which works out at roughly one minute per question. Most people finish with time to spare.

How accurate is an online IQ test?

A well-built one is useful but not clinical. This test reports a 95% confidence interval of roughly ±9 points near the middle of the range, widening at the extremes. A supervised test such as the WAIS-IV is considerably more precise, and only a qualified psychologist can produce a score with clinical standing. The full comparison is here.

What is a good IQ score?

By construction the average is 100 and about two thirds of people score between 85 and 115. A score near 130 is around the 98th percentile. Because scores are defined relative to other people rather than against an absolute standard, "good" only means "compared with the norm group". See the full range table.

Do you store my answers or my score?

No. The test runs entirely in your browser. Your answers are held in your own browser's session storage so you can refresh without losing progress, and they are discarded when you close the tab. Nothing is uploaded to us, because there is no server to upload it to.

Can I retake the test?

You can, but a second attempt on the same questions will overestimate you: you have already seen the items and worked out some of the rules. This is called a practice effect, and it is why real test batteries keep alternate forms and recommended retest intervals.

Ready to start?

Thirty questions. One score, with the uncertainty attached.

Start the test

30 questions · about 30 minutes · no sign-up