Are Online IQ Tests Accurate? What the Research Actually Shows
The short answer: some online IQ tests are accurate enough to give you a rough estimate, and plenty are not accurate at all.
A carefully built online test can place you in roughly the right range. No online test, ours included, can give you a number to rely on for school, work, Mensa or a diagnosis.
The difference between a useful test and a useless one comes down to a few things you can check yourself. This guide shows you what they are.
Key points
- Across 49 studies, people scored about 3 IQ points higher on unsupervised online ability tests than on supervised ones.
- Taking a test a second time adds about 5 points on average, so your first attempt is usually the most honest.
- The weakest part of most online tests is who they compare you with, not the fact that they run in a browser.
- Mensa, schools and clinicians only accept tests taken under supervised, standard conditions.
What "accurate" means for an IQ test
Psychologists judge an intelligence test on three separate qualities, and a test can pass one while failing another.
Reliability asks whether you would get a similar score if you took the test again under the same conditions. Validity asks whether the test measures reasoning ability rather than something else, such as reading speed or how many puzzle apps you have played. Norming asks who your score is being compared with.
That last one matters more than most people realise. An IQ score is not a raw mark out of 100. It is a comparison: 100 is the average for a reference group, and every 15 points is one standard deviation away from that average. Our guide to what counts as a good IQ score shows the full chart. If the reference group is wrong, the score is wrong, however well written the questions are.
How a professional IQ test is built
It helps to see what online tests are being measured against. The best-known adult test is the Wechsler Adult Intelligence Scale. Its fifth edition, the WAIS-5, was released by Pearson in August 2024 and covers people aged 16 to 90.
A trained examiner gives it one-to-one, and the core full scale IQ takes about 45 minutes. Pearson re-standardises its cognitive tests roughly once a decade, because the population's scores shift over time. A test normed on people from 30 years ago will make today's test-takers look cleverer than they are.
How much do norms drift? Less than they used to. When researchers gave the same people both the WAIS-IV and the WAIS-5, the expected gap of about 3 points turned out to be roughly 1.2 points. That is small, but it shows why the age of a test's norms is worth asking about.
Even with all this care, a psychologist's report gives a range around the score rather than a single exact number. If the gold standard needs a margin of error, any online score needs a bigger one.
One thing that is not the problem is the screen. In Pearson's own study of the WAIS-5, the digital format did not account for score differences compared with pen and paper. The issues with online tests lie elsewhere.
Why online IQ scores drift away from reality
The comparison group is usually wrong
Professional tests are normed on large samples chosen to match the population by age, education and other factors. Most online tests compare you with whoever else happened to take the test on that website. People who seek out IQ tests are a self-selected group, not a cross-section of the country, so where you rank among them says little about where you rank in the population.
Many sites never say who they compare you with at all. If you cannot find that information, the score has nothing solid underneath it.
Nobody is watching
At home you can pause, look things up, ask someone or retake a tricky section. A 2020 meta-analysis by Steger, Schroeders and Gnambs pooled 49 studies with more than 100,000 people and found that unsupervised ability tests produced scores about 0.2 standard deviations higher than supervised ones.
The same study found the gap shrank for tasks that are hard to search for online. That is good news for reasoning tests: you cannot easily Google the answer to an abstract pattern puzzle, while a general-knowledge question takes seconds to look up.
The second attempt is inflated
People get better at a test simply by having taken it before. A large meta-analysis by Scharfen, Peters and Holling (2018) found that scores rose by about a third of a standard deviation, around 5 IQ points, from the first attempt to the second. By the third attempt the gain reached about half a standard deviation, and then it levelled off.
None of that reflects a real change in ability. You have learnt the question styles, the timing and the traps. That is why your first honest attempt at any test is the one to take most seriously.
Short tests only measure a slice
A professional battery combines several kinds of task: verbal reasoning, working memory, processing speed, visual puzzles and more. A ten-question quiz built entirely from one puzzle type can tell you how you did on that puzzle type, and not much beyond it.
This is also why single-skill tools, such as our Number Memory Test or Reaction Time Test, are presented as checks on one ability rather than as IQ tests.
Some tests are designed to flatter
A properly normed test has an average of 100 by definition, so about half of the people who take it should score below 100. If almost everyone you know scores above 120 on the same site, that is a sign the scale is inflated. We look at this pattern in what an IQ of 120 really means.
Flattering scores are good for business, especially when a certificate or "full report" sits behind a payment screen. The same instinct explains the famous IQ figures quoted for historical figures, many of which were never measured. Our article on Einstein's supposed IQ of 160 traces one of them.
Where online testing holds up
None of this means every browser-based test is worthless. The International Cognitive Ability Resource (ICAR), developed by Condon and Revelle at Northwestern University, is a set of public-domain reasoning items built for online research.
When it was tested on a university sample and an online sample of around 97,000 people, the researchers found that the items remained valid despite being freely available. Carefully built and properly checked online reasoning tests can produce meaningful results. What separates them from the rest is design, a clear comparison group and honest reporting.
Professional test vs online test: side by side
| Supervised professional test | Well-built online test | Low-quality online test | |
|---|---|---|---|
| Who runs it | A trained examiner, one-to-one | You, on your own | You, on your own |
| Compared with | A large sample matched to the population | A stated, described group | Unknown or never mentioned |
| Skills covered | Several, including verbal, memory and speed | Usually a few reasoning areas | Often one puzzle type |
| What you get | A score with a confidence range and a skill profile | An estimate with a clear explanation | A single, often flattering number |
| Cost | A professional fee | Free or low cost | "Free", then a payment screen |
| Accepted officially | Yes, if it is an approved test | No | No |
Check any online IQ test in 30 seconds
Tap each statement that is true of the test you took. Nothing is saved or sent anywhere.
What Mensa accepts
Mensa International sets its entry bar at the top 2% of the population, which is around 130 on the standard scale. Our guide to what an IQ of 130 means explains how rare that is.
The score must come from an approved intelligence test that was properly administered and supervised. Mensa states plainly that online tests cannot be used for admission, and that even its own online IQ Challenge is for practice only. Schools, employers and clinicians take the same view for the same reasons.
How to get the fairest result from an online IQ test
- Take it rested and in one sitting. Tiredness and interruptions lower scores for reasons that have nothing to do with reasoning ability.
- Use a screen big enough to see the puzzles clearly. Squinting at a small pattern on a phone adds errors of its own.
- Don't look anything up or ask for help. It only inflates a number that nobody else will see.
- Treat your first attempt as your real result. Later attempts carry a practice boost of several points.
- Read the score as a range, not a point. Think "somewhere around 110", not "exactly 112".
- Look for a method page. A test that explains its scoring deserves more trust than one that hides it.
How accurate is the IQScore test?
We would rather you trusted our number a little less than trusted it too much, so here is the honest version.
The IQScore cognitive assessment asks 36 reasoning questions across four areas: pattern, number, logical and applied reasoning. You have 25 minutes, and the questions and answer options are shuffled on every attempt. Reasoning puzzles like these are exactly the kind of task the research found hardest to game online.
It has not been standardised on a representative national sample the way the WAIS-5 has, and it does not measure verbal knowledge, memory or processing speed. Your result is an estimate on a scale of 70 to 130, and the result page explains how it is worked out.
What we can promise is the part many sites skip: no email, no sign-up and no payment screen standing between you and your score. It is best used for curiosity, practice and tracking your own progress. Our guide to free IQ tests with instant results covers what else to expect.
When you need a professional assessment
An online score is fine for satisfying curiosity. It is not the right tool when something real depends on the answer. That includes questions about learning difficulties, entry to gifted or specialist programmes, workplace adjustments, clinical concerns or legal matters.
In those cases, speak to a qualified psychologist who can give a standardised test in person and interpret it alongside the rest of your history. The same applies if an online result, or your everyday experience of memory and attention, is worrying you.
Frequently asked questions
Are online IQ tests accurate?
Some give a reasonable estimate of your reasoning ability, especially well-designed tests that explain their scoring. Many do not. No online test is accurate enough to replace a supervised assessment by a qualified psychologist.
Why did I score higher online than on a supervised IQ test?
Three common reasons: unsupervised tests score about 3 points higher on average, repeat attempts add around 5 points through practice, and many online tests compare you with a flattering or unknown group rather than the general population.
Can I use an online IQ score to join Mensa?
No. Mensa only accepts scores from approved intelligence tests that were properly administered and supervised. Its own online IQ Challenge is for practice only.
Is my first or second attempt more accurate?
Usually your first. Research on retesting shows scores rise by about 5 IQ points on a second attempt because of familiarity with the test, not because ability has changed.
Are paid IQ tests more accurate than free ones?
Not necessarily. Paying for a certificate or report after the test does not make the underlying score more accurate. What matters is how the test was built, who it compares you with and whether it explains its method.
How accurate is the IQScore test?
It gives an estimate based on 36 reasoning questions. It has not been standardised on a representative national sample, so treat the result as a rough guide for curiosity and practice rather than an official IQ score.
Keep reading
- What Is a Good IQ Score?The full IQ chart, with percentiles for every score band.
- What Does an IQ of 120 Mean?How rare 120 really is, and why online tests hand it out so often.
- What Does an IQ of 130 Mean?The score that sits at Mensa's threshold, explained.
- What Was Einstein's IQ?Where the famous 160 figure came from, and why it was never measured.
- Free IQ Test with Instant ResultsWhat a no-email, no-sign-up test can and cannot tell you.
Sources
- Steger, D., Schroeders, U. and Gnambs, T. (2020). A meta-analysis of test scores in proctored and unproctored ability assessments. European Journal of Psychological Assessment, 36(1), 174–184.
- Scharfen, J., Peters, J. M. and Holling, H. (2018). Retest effects in cognitive ability tests: a meta-analysis. Intelligence, 67, 44–66.
- Condon, D. M. and Revelle, W. (2014). The International Cognitive Ability Resource: development and initial validation of a public-domain measure. Intelligence, 43, 52–64.
- Pearson Assessments. Overview of the WAIS-5, and WAIS-5 digital and paper format equivalence study (2024).
- Healthcare Dive (2024). The release of the WAIS-5.
- Journal of Intelligence (2024). Wait, where's the Flynn effect on the WAIS-5?
- Mensa International. Getting your IQ tested: FAQs.
IQScore tests are for education and entertainment. Scores are estimates and are not a clinical or diagnostic result. If you have concerns about memory, attention or learning, please speak to a qualified professional.
