Type "IQ test" into a search bar and you will find hundreds of free quizzes that promise a number in five minutes. Some are fun and well made. None of them can tell you your IQ in the sense a psychologist means it. This guide explains why, what a score does and does not mean, and how to get the most out of an online test anyway.
What an IQ score actually is
An IQ score is not a measurement like height. It is a comparison: how your performance on a set of tasks stacks up against a reference group, usually people of your own age. Modern IQ scales are set so the average score is 100 and one standard deviation is 15 points (OpenStax Psychology). Major clinical tests use this scale; the Stanford-Binet Fifth Edition, for example, reports its composite IQ scores with a mean of 100 and a standard deviation of 15 (Stoelting).
Two ideas hold this together:
- Standardization means everyone gets the same instructions, the same materials and the same scoring rules.
- Norming means the test was given to a large, representative sample so that raw results can be turned into meaningful scores. OpenStax notes that a representative sample is what lets the bell curve describe the wider population.
Without the second step, a "100" is just a number the website chose.
How a professional test differs
The Wechsler Adult Intelligence Scale, Fifth Edition (WAIS-5), published by Pearson, covers ages 16 through 90. Pearson states that its standardization sample was collected in 2023-2024 and matches current census data on five demographic variables. The full-scale score is built from seven subtests and takes roughly 45 minutes; a longer version with ten subtests takes about an hour (Pearson). The Stanford-Binet Fifth Edition was normed on 4,800 people from age 2 to 85 and older (Stoelting).
Several things set these apart from a web quiz:
- A trained examiner. The American Psychological Association says licensed psychologists are trained to give assessments and interpret the results, and a full evaluation can also include interviews, records and observation (APA).
- A controlled setting. Timing, instructions and scoring are the same for everyone. A quiz taken on a phone on the bus is not.
- A reported margin of error. Pearson's sample WAIS-5 report lists a full-scale score of 111 with a 95% confidence interval of 106 to 116 (Pearson sample report). Even the gold standard produces a range, not a single exact number.
Why free online tests are not normed
Building a norm sample means recruiting thousands of people who reflect the population by age, region, education and more, then testing them under controlled conditions. That is slow and expensive work, and a free quiz does not do it. This is our reasoning rather than a published statistic, but it follows directly from how norming is defined: a site whose visitors are self-selected, take the test on any device, and may repeat it cannot say how a score compares with a representative group.
Many sites still print a confident "Your IQ is 128." What that number usually reflects is the site's own scoring formula. Treat it like a personality quiz result: interesting, not diagnostic.
Our Free IQ Test is for entertainment. It is a set of puzzles to enjoy and compare with friends, not a clinical instrument, and it should not be used for school, work, medical or legal decisions.
The Flynn effect: scores drift over time
Raw performance on IQ tests has risen for generations, a pattern called the Flynn effect. A meta-analysis by Pietschnig and Voracek covered 271 independent samples, almost 4 million participants and 31 countries between 1909 and 2013. It found average gains of about 0.28 IQ points per year on full-scale scores, with larger gains in fluid (reasoning) ability, about 0.41 per year, and smaller ones in crystallized (knowledge-based) ability, about 0.21 per year (APS). That works out to roughly three points per decade. The authors also report that gains have weakened substantially in more recent decades (EurekAlert summary).
Why this matters for you: tests must be re-normed periodically, which is one reason the WAIS-5 has a fresh 2023-2024 sample. An old test scored against old norms, or an online quiz with no norms at all, tends to give flattering numbers. Researchers disagree about causes; the authors name better nutrition, health care and education as leading candidates and suggest the gains reflect specific abilities and test-taking strategies more than general brainpower.
Practice effects: the second try is easier
If you take similar tests more than once, you improve partly because you have learned the format. A meta-analysis of 122 studies and 153,185 participants by Scharfen, Peters and Holling found significant retest effects, with no further gains after the third administration. In their data, the average gain after one retest was about 0.33 standard deviations and about 0.50 after three tests; the gain was larger when the identical test was repeated (about 0.37) than when an alternate form was used (about 0.23).
The study measures gains in standard deviations of the first test. On an IQ scale with an SD of 15, that is roughly 5 points after one retest and 7 points after three. This conversion is our own arithmetic, and the study covered many kinds of cognitive tests rather than one specific IQ test, so treat it as a rough illustration.
The APA's own advice for clinical assessments runs the other way: it recommends not practicing or studying for them, since that can distort the picture (APA). If you take several online tests, your later scores say as much about familiarity as about ability.
How to read a score: the distribution table
If IQ scores follow a normal curve with a mean of 100 and an SD of 15, we can work out what share of people fall below any score. Every percentage in the table below is computed by us from the normal distribution, not taken from a test publisher. Real score distributions are only approximately normal, especially at the extremes, so treat the tails as estimates.
| Score | Distance from mean | Share scoring below (computed) | Share scoring above (computed) |
|---|---|---|---|
| 70 | -2 SD | 2.3% | 97.7% |
| 85 | -1 SD | 15.9% | 84.1% |
| 100 | mean | 50.0% | 50.0% |
| 115 | +1 SD | 84.1% | 15.9% |
| 130 | +2 SD | 97.7% | 2.3% |
| 145 | +3 SD | 99.87% | 0.13% |
Sourced cross-checks: OpenStax states that about 68% of people score between 85 and 115, that about 2% score 130 or above, and that about 2.2% score below 70. These match our computed figures (68.3%, 2.3% and 2.3%). Pearson's sample report lists a full-scale score of 111 at the 77th percentile, which also agrees with the normal curve (about 76.8%).
A practical example: suppose a quiz says you scored 130. Because a single result carries error, an unnormed quiz has no known margin, and you may have practiced similar puzzles, the safe reading is "I did well on these puzzles," not "I am in the top 2% of the population."
What IQ tests cannot tell you
- Everything about you. They do not measure creativity, curiosity, persistence, social skill or practical judgment. OpenStax notes that how valid IQ tests are remains a matter of debate.
- Your exact number. Even professional scores come with a confidence interval, as the WAIS-5 sample report shows.
- Your worth or potential. A score describes performance on a given day on a given set of tasks.
If you enjoy mental challenges for their own sake, brain teasers and math trivia are a more honest way to use that curiosity. To learn about your preferences and style rather than your reasoning, try our personality test.
Where to get a real assessment
If you need a valid score, for example for a school accommodation, a clinical question or a gifted-program application, ask a licensed psychologist. The APA describes licensed clinical psychologists as the professionals trained to administer and interpret formal assessments (APA Help Center). Practical steps (our recommendation):
- Start with the reason you want testing, so the psychologist can pick the right instrument and report format.
- Ask your doctor, your child's school psychologist, or a state or provincial psychological association for referrals.
- Confirm the clinician's license and ask which test they use and whether it is the current edition.
- Do not rehearse on practice versions beforehand.
Quick answers
Is an online IQ test ever accurate?
It can be a reasonable puzzle sampler, but without norming and a controlled setting it cannot give you a score comparable to a professional test.
Should I retake a test to get a higher score?
You probably will score higher, but according to the retest research that gain reflects familiarity, not a change in ability.
Is 100 a good score?
By design it is the average, so half of the reference group scores higher and half lower.
Want to try something playful? Take our Free IQ Test for entertainment, and keep the score in perspective.
Related quizzes
Sources
- Pietschnig & Voracek (2015), One Century of Global IQ Gains, Perspectives on Psychological Science
- EurekAlert: Flynn effect meta-analysis press summary
- Scharfen, Peters & Holling (2018), Retest effects in cognitive ability tests, Intelligence
- Pearson Assessments: WAIS-5 overview brochure
- Pearson Assessments: WAIS-5 sample score report
- Stoelting: Stanford-Binet Intelligence Scales, Fifth Edition (SB5)
- American Psychological Association: Psychological assessment (Help Center)
- OpenStax Psychology 2e: Measures of Intelligence