Free IQ test with instant results - no sign-up

What a free IQ test actually costs you

IQTest costs nothing at all — no email field, no card field, no paid tier — and the score, percentile, error range and five-index profile appear together the moment you finish. Search “free IQ test” and most of what comes back is free in the sense that starting is free. The gate arrives at the end, after you have already spent twenty minutes and feel invested. There are four common versions of that gate, and it is worth knowing which one you are walking into before you begin. We have put ourselves in the table too, in the row we actually belong in.

IQTest sits in none of those four rows — easy to claim, easy to check: reach the end of the test and see whether anything is asked of you. The rows are not equivalent to each other either, and the bottom one is documented rather than hypothetical. Reviewers on Trustpilot describing one such service report paying a token amount — fifty pence, a dollar, a single euro — to see a score, then finding monthly charges in the region of €14.99 to $29.99 appearing weeks later. Several say they cancelled the card to make it stop. Complaints with that shape were still being posted in January 2026.

What free means here, item by item

• No card field exists on this site. Nowhere, at any point. There is no paid tier to upgrade to, so there is nothing to bill you for.
• No email field either. You are not asked who you are at any point, so there is no address to add to a list and no result waiting to be emailed to you.
• Nothing is withheld. Score, percentile, confidence range, five-index profile and the reading list all appear together the moment you finish. There is no second step.
• Scoring happens on your device. The arithmetic runs in your browser, which is the mechanical reason none of the above is possible: there is no server holding a result that could be gated. See the privacy policy for what that means in practice.

What the test measures

IQTest measures five abilities separately and reports each one. A single composite averages them, and the profile underneath is usually the more useful half of the result.

Twenty-one items and two tasks answered alone at a laptop is a screener, not a diagnosis, and most of this page is about where a screener stops being useful.

Scoring does not simply count correct answers. Every item carries a difficulty value, and your result is the ability level that best explains the whole pattern of responses — clearing a hard item moves it further than clearing an easy one, and a built-in guessing correction means blind answering earns nothing. Your margin of error is calculated for you as well — answering consistently with each item’s difficulty narrows it, a scattered pattern widens it — rather than printing the same interval for everybody. The full build and scoring notes give the model, the item parameters and the defects that were fixed. The result leads with your profile rather than the number, because the profile is the part that does not depend on norms — which of your own abilities is stronger is a comparison against yourself.

An earlier version of IQTest left working memory and processing speed unmeasured, arguing that a browser cannot present items at a controlled rate. That was half right: a self-paced form cannot measure memory span, but forced pacing in a browser works fine, and forced pacing is the part that matters. Both are measured now, and the methodology page corrects the earlier claim rather than quietly dropping it.

Your result is a range, not a number

An IQTest result is reported as a band because the underlying measurement carries error, and the quantity that describes that error is the standard error of measurement. Properly normed tests publish a confidence interval instead of a bare figure for the same reason. The arithmetic is not complicated: a 95% interval is the observed score plus and minus 1.96 × SEM. With a SEM of 3, that is about six points either way.

Consider the benchmark. Full-scale reliability on the Wechsler Adult Intelligence Scale sits near .98, which is exceptionally high for a psychological instrument, and even there Pearson’s own score reports print a confidence band around the FSIQ rather than a single number. If the best-validated adult IQ test in the world declines to state a point estimate, a twenty-one-item web screener with two brief tasks certainly should not.

This matters most at boundaries. People chase a 130 because of what it unlocks socially, then treat 128 as a failure and 131 as an achievement. Those two results are indistinguishable from measurement noise alone.

Score bands and the percentiles behind them

On the deviation-IQ scale the population mean is fixed at 100 and the standard deviation at 15. Percentiles follow from the normal curve. The right-hand column is there because raw percentages are hard to feel.

Band labels come from clinical convention and carry no moral content. A score of 95 sits in the widest, most crowded part of the distribution, alongside half the adult population. Our guide to reading a score goes further: why percentile is not what most people assume, and how the same position produces different numbers on different scales.

How far an unsupervised score can be trusted

An IQTest result will usually place you in the correct 10-point band and can miss by more than that near the edges of one. Four things drive the gap between a web result and a psychologist’s report, and each pushes in a knowable direction.

1. The comparison group is self-selected. Clinical norms come from a sample matched to census demographics. Everyone who takes an online IQ test chose to, which is not the same population and never will be.
2. Nobody is watching. You can pause, look something up, retry a question, or take it at 2 a.m. after a long shift. Those cut in both directions and the score cannot tell which happened.
3. Twenty-one items and two brief tasks are a small sample of a large ability. The WAIS-5 runs about forty-five minutes for a full-scale score for exactly this reason.
4. Familiarity inflates results. Matrix and rotation puzzles are learnable as formats. Someone on their fifth online IQ test is partly measuring their practice with the genre.

A reasonable expectation: this test will usually place you in the right band and can be off by more than you would like near the edges of it. That is the honest ceiling for the format, and any free web test claiming clinical precision is overselling.

There is a second uncertainty underneath that one, and almost nobody states it. Measurement error describes how much a score bounces on a retake. Norm error describes how far the whole scale might be displaced, and it exists whenever a test is anchored to published population parameters rather than to a standardisation sample the publisher collected. Ours is anchored that way, so the range we print is a floor on the true uncertainty rather than the whole of it. The methodology page publishes what has and has not been established — currently none of the five things a psychometric instrument needs, which is the same position as every free test in this market and a different position on saying so.

Mensa eligibility: what an IQTest score can and cannot do

Mensa admits at the 98th percentile on an approved supervised test. Because different instruments use different standard deviations, the same percentile wears different numbers:

Those are one threshold written three ways, not three difficulty levels. No online test qualifies you, ours included — Mensa accepts only its own supervised admission test or prior evidence from an approved proctored instrument. American Mensa charges $60 for its supervised test and requires candidates to be at least 14.

Where a screener does help is deciding whether to bother. If you land around 105 here, paying $60 to be told you are near the middle is an expensive confirmation. If you land at 128 or above, the supervised test becomes a reasonable use of an afternoon.

Where a supervised, professionally scored test comes from

For adults the standard instrument is the Wechsler Adult Intelligence Scale. Pearson released the fifth edition, WAIS-5, in late 2024, replacing a WAIS-IV that had been the clinical default since 2008. The revision refreshed norms to a 2023–24 sample, added subtests, and split the old Perceptual Reasoning Index so that fluid reasoning and visual-spatial ability are now reported separately.

It is administered one-to-one by a qualified psychologist and cannot be bought or self-scored; Pearson restricts purchase by qualification level. Practically, that means an appointment, a couple of hours, and a written report. Fees vary by country and by whether the assessment is standalone or part of a broader evaluation, so ask for the total in writing before booking.

Can you raise your IQ score?

You can raise an IQTest score by retaking the test — a second sitting typically returns a higher number on the same ability, because a familiar format stops being novel. Whether the underlying ability moved is a separate question, and the research is far less encouraging than the brain-training market suggests.

Practice effect is the gain from having met the format before; it is real, it appears on a second sitting, and it fades as the novelty of the item type wears off. Transfer is improvement on the underlying ability that shows up on tasks you never trained. Transfer is the contested one, and the burden of proof sits with anyone selling it.

For a cleaner reading, take the test once, awake, on a device you like, without interruptions. Do not take it four times in a week and quote the highest result — that number is measuring your memory of the questions.

US reasoning scores have been drifting since 2006

Composite reasoning scores among US adults fell between 2006 and 2018. Dworak, Revelle and Condon published the analysis in Intelligence in 2023, covering 394,378 adults tested through the SAPA Project on ICAR items. Composite ability scores declined across the period. So did matrix reasoning and letter-and-number series. Verbal reasoning slipped without reaching their threshold for a meaningful slope.

One domain moved the other way: three-dimensional rotation rose from 2011 to 2018. The declines held regardless of age, education or gender, and were steepest among 18–22 year-olds and those with less formal education.

Lead author Elizabeth Dworak has been explicit that the result does not show Americans becoming less intelligent — it may reflect people getting worse at taking these particular kinds of test. Paraphrased from Dworak’s comments to Northwestern University, March 2023

Two practical consequences. First, norms age, and a test calibrated on a 2008 sample is scoring you against people who are no longer the reference population — part of why Pearson re-standardises roughly every decade. Second, if your matrix-reasoning subscore looks lower than you expected next to your spatial subscore, you are pointed in the same direction as a population-level trend, not necessarily at a personal deficit.

Questions about IQTest and its scoring

What an IQ score does not measure

An IQ figure describes performance on a set of reasoning puzzles relative to a comparison group. It is not a measure of worth, creativity, judgement, kindness or how well anyone will do in a career. It says nothing about learning disabilities, ADHD, autism or giftedness, and it must never be used to self-diagnose any of them.

If you are worried about a real cognitive change — memory, concentration, word-finding, or a difference from how you used to function — a web quiz is the wrong tool and a clinician is the right one. IQTest is an educational site and is not affiliated with Mensa, Pearson or any test publisher named on this page.

Two related pages spell this out further. Our editorial position lists the five things we have decided never to publish, celebrity IQ figures among them. The conditions of use name the specific purposes a score from this site is not valid for, including Mensa applications, diagnoses and hiring decisions.

Sources

• 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
• Dworak, E. M., Revelle, W. & Condon, D. M. (2023). Looking for Flynn effects in a recent online U.S. adult sample. Intelligence, 98, 101734. doi:10.1016/j.intell.2023.101734
• Pearson Clinical Assessment — Wechsler Adult Intelligence Scale, Fifth Edition (WAIS-5) product documentation, released 2024. Checked 22 July 2026.
• American Mensa — published qualifying criteria and supervised testing fee. Checked 22 July 2026.
• Trustpilot — public reviews of a paid online IQ report service, posts dated through January 2026, used as attributed user testimony rather than verified fact.

Every factual claim on this page was checked against the source above on 22 July 2026. If something here is out of date or wrong, tell us and we will correct it with a note.