Are online IQ tests accurate? How to tell a real test from a scam
Search for “IQ test” and you get two kinds of results. One is Mensa’s practice puzzles, which say upfront that the score is just for fun. The other is a long list of sites that let you solve thirty puzzles and then ask for your card number before they show the result. Real tests exist somewhere in between, but they are rare, and the reasons have more to do with money than with science.
What an IQ score is
An IQ score is a rank, not a measurement like height or weight. A test counts your correct answers and compares that count with a norm group: a large number of people who took the same items under the same conditions. The scale is set so that the average of that group is 100 and the standard deviation is 15. A score of 115 means you did better than about 84% of the norm group. A score of 130 means better than about 98%.
That has a simple consequence. A test without a norm group cannot give you an IQ at all. It can give you a count of correct answers, and anything on top of that is a number somebody made up.
Why so many tests use matrices
The classic format is Raven’s Progressive Matrices, published by the British psychologist John C. Raven in 1938. You see a 3×3 grid of shapes with the last cell missing and pick the piece that completes the pattern. There are no words and no arithmetic, so the result depends less on schooling and language than with most other tests. That is why matrices became the standard way to measure fluid reasoning, the ability to spot a rule you have never seen before.
The rules behind a matrix are simple: a shape rotates, the number of dots grows by one in each row, a fill changes from empty to striped to solid. Simple rules can be generated by software. In 2010 researchers at Sandia National Laboratories published a program that builds Raven-style matrices, together with norms for the items it produced. Psychologists have also released free items for research, such as the International Cognitive Ability Resource (ICAR), which includes a set of matrices.
So why is a good free test so hard to find?
There are three reasons, and none of them is technical.
The famous test can’t be online. Raven’s matrices are a commercial product sold to qualified professionals. Psychologists are bound by an ethics code that tells them to protect test materials, so the real items are kept off the internet on purpose. Anything online that calls itself “the Raven test” is at best a copy of the format.
Items are cheap, norms are expensive. Generating a thousand matrices takes an afternoon. Finding out how hard each one is takes thousands of people solving them under the same conditions. Researchers do that work for papers, not for a website, and they don’t buy search ads.
Scams pay better. Truth in Advertising described a typical case in March 2026: a “free” six-minute IQ test advertised on Google, a $1 charge to see the score, and fine print that signs you up for $30 a month unless you cancel within seven days. A site that earns $30 a month from one visitor can outbid anyone for ads on “IQ test”. An honest test that earns nothing cannot.
Mensa’s free online challenge is the honest exception near the top: 35 puzzles in 25 minutes, with a note that the result is for entertainment only and won’t qualify you for membership. Mensa admits members on the basis of supervised tests.
Seven checks before you trust a score
- You see the score without paying or leaving an email. A paywall after the last question is the most common sign of a trap.
- The test tells you who you are compared with, and how many people are in that group. “Our scientists” is not a norm group.
- Everyone gets the same conditions. The same items, or items of known difficulty, and the same time limit. Scores from different conditions can’t be compared.
- It shows how precise the score is. A short test measures with an error of several IQ points. A result like “IQ 127” with no range claims more precision than any 20-minute test has.
- Retakes are marked as retakes. People do better the second time they take the same kind of test. A meta-analysis of 122 studies found an average gain of about a third of a standard deviation on the second attempt, roughly 5 IQ points, and half a standard deviation by the third. A test that counts retakes in its norms inflates them.
- Not everyone comes out a genius. On a real IQ scale about 2% of people score 130 or more, and half score below 100. If all your friends got 125, the scale is flattering you.
- It doesn’t sell a certificate. No online test can certify your IQ. That takes a psychologist, a standardized test and a supervised session.
What a free online test can honestly tell you
It can tell you how well you reason with matrices compared with the other people who took the same test, on a given day, on a screen. That is a narrower claim than “your intelligence”, and a useful one.
Norms also age. In the 20th century scores on IQ tests rose by about 3 points per decade, which is known as the Flynn effect. In Norway the trend reversed for men born after the mid-1970s. A test normed decades ago flatters or punishes you depending on which way the population has moved since.
How our test does it
Pattern is our 20-minute matrix test. Each sheet has 20 matrices from our own generator, and they get harder as you go. Your score is your rank among the first attempts of everyone who took the test. It appears on the IQ scale once 50 people have done the test, and once there are enough results you also see the reliability of the test and a 90% range around your score. Retakes get a score but stay out of the norms. No sign-up, no email, no payment.
The honest limit: our norm group is the people who play on this site, not a random sample of the population. People who take IQ tests online are probably not average, so 100 means the average player, not the average human.
Sources
- Raven, J. C. (1938). Progressive Matrices: A Perceptual Test of Intelligence. London: H. K. Lewis.
- Matzen, L. E., Benz, Z. O., Dixon, K. R., Posey, J., Kroger, J. K., & Speed, A. E. (2010). Recreating Raven’s: Software for systematically generating large numbers of Raven-like matrix problems with normed properties. Behavior Research Methods, 42(2), 525-541. doi:10.3758/BRM.42.2.525
- Condon, D. M., & Revelle, W. (2014). The International Cognitive Ability Resource: Development and initial validation of a public-domain measure. Intelligence, 43, 52-64. doi:10.1016/j.intell.2014.01.004
- American Psychological Association. Ethical Principles of Psychologists and Code of Conduct, standard 9.11: Maintaining test security. apa.org/ethics/code
- Truth in Advertising (2026, March 16). Brainowl’s “free” IQ test. truthinadvertising.org
- Mensa International. Mensa IQ Challenge. mensa.org
- Scharfen, J., Peters, J. M., & Holling, H. (2018). Retest effects in cognitive ability tests: A meta-analysis. Intelligence, 67, 44-66. doi:10.1016/j.intell.2018.01.003
- Flynn, J. R. (1987). Massive IQ gains in 14 nations: What IQ tests really measure. Psychological Bulletin, 101(2), 171-191. doi:10.1037/0033-2909.101.2.171
- Bratsberg, B., & Rogeberg, O. (2018). Flynn effect and its reversal are both environmentally caused. Proceedings of the National Academy of Sciences, 115(26), 6674-6678. doi:10.1073/pnas.1718793115