Looking for the best IQ tests of 2026? This British English guide ranks 12 trusted online IQ tests by accuracy, time, cost and certificate options, with the psychometric science explained in plain English. Every test below has been used, timed and assessed against six evaluation criteria, and every recommendation links straight to the platform so you can begin your test in seconds. Whether you want a serious score for personal development, a Mensa-style practice run, or a certified result you can print and share, you will find a suitable option on this list.
Online intelligence tests vary enormously in quality. Some are built on properly normed item banks with published psychometric data; others are little more than gamified quizzes dressed up in scientific language. To separate the credible from the cosmetic, we applied a transparent six-factor framework drawn from the APA Standards for Educational and Psychological Testing (2014), which remains the global benchmark for assessment validity.
Construct validity. Does the test measure what it claims to measure, namely general cognitive ability (often called g)? We looked for tests that combine verbal, numerical, spatial and abstract items, since the strongest indicator of general intelligence is performance across multiple unrelated domains.
Test-retest reliability. A reliable test produces consistent scores when the same person retakes it after a reasonable interval. Properly normed online tests typically achieve a reliability coefficient between 0.70 and 0.85. Clinical instruments such as the WAIS-5 sit much higher, with index reliability coefficients between 0.90 and 0.97 reported in the WAIS-5 Technical and Interpretive Manual (Wechsler et al., 2024).
Norming sample transparency. Trustworthy tests publish the size and demographics of the population against which your score is compared. If a platform cannot describe its norming sample, the percentile it returns is essentially decorative.
Scoring method. We favoured tests using a Wechsler-style standardised scale, where 100 is the population mean and 15 points equal one standard deviation. We flagged tests that use unusual scales without explanation.
User experience and accessibility. Time pressure, clear instructions, a working timer, no distracting advertisements during the test, and a mobile-friendly interface all influence how reliably your true ability is captured.
Honest reporting. A trustworthy report includes a percentile, a standard error of measurement, a brief explanation of what the score does and does not say about you, and ideally a breakdown by cognitive domain.
We excluded tests that promise a specific IQ score in under five minutes, those that paywall the result without showing a preview, those with no time limit at all, and those that asked for an email address before showing even a sample question. None of these practices is consistent with mainstream psychometric standards.
Visit: officialiqtests.com
Official IQ Tests earns the top slot in 2026 for its balance of scientific seriousness and everyday usability. The 30-question battery covers verbal analogies, numerical sequences, spatial rotation, pattern recognition and short working-memory tasks, giving a genuine sampling across the five Wechsler index domains rather than focusing on a single reasoning style. The test takes around 20 minutes to complete and returns a Wechsler-style score on the 100-mean, 15-SD scale.
What sets this platform apart is the post-test reporting. Rather than handing you a bare number, the results page shows your score against the bell curve, your approximate percentile, and a sensible note on the standard error of measurement. There is no email gate before you can see your result. For anyone wanting an honest first read on their cognitive profile, this is the test we recommend by default.
Visit: iqcertificate.org
If you need something you can frame, attach to a CV or send to a recruiter, IQ Certificate is the cleanest option of the year. The 30-question assessment completes in under 30 minutes and ends with a downloadable, dated certificate displaying your score in standard Wechsler format. Items rotate across verbal, mathematical and visual-spatial reasoning, with the timer encouraging the kind of pace you would experience under supervised conditions.
The certificate itself is the obvious draw, but the underlying test is solid. We particularly liked that the platform explains its scoring method on the same page and does not hide the basics behind a sign-up wall. The credential is naturally most useful for personal motivation, scholarship applications and high-IQ society practice rather than for clinical or employment purposes, where only supervised testing carries weight.
Visit: actualiqtest.com
Culture-fair tests use non-verbal items such as matrices, sequences and shape rotations, so that vocabulary and cultural background influence the score as little as possible. Actual IQ Test takes this principle seriously. The 25-question set is dominated by visual-pattern problems modelled on Raven's Progressive Matrices, with a few numerical sequences thrown in for balance.
The result is a test that travels well across languages and educational backgrounds, which is one reason we recommend it for users whose first language is not English. Completion takes around 15 minutes, the interface is uncluttered, and the report places your score on a bell curve with a clear percentile band. If you suspect verbal items have penalised your score on previous tests, start here.
Visit: realiqtest.net
Real IQ Test blends classical IQ items with newer cognitive challenges that reflect current research into fluid reasoning. The 30-question battery is genuinely varied: you will see verbal analogies in the Cattell tradition, matrix problems in the Raven's tradition, and number-pattern items closer to what appears on the WAIS-5 Figure Weights subtest. Completion takes roughly 20 minutes.
The platform publishes a counter of completed assessments running into the hundreds of thousands, which means scores are normed against a meaningful real-world distribution rather than a small theoretical sample. The post-test breakdown is detailed enough to be useful for self-development, showing how you performed across reasoning categories rather than collapsing everything into a single figure.
Visit: legitimateiqtest.com
The name is a touch defensive, but the substance is sound. Legitimate IQ Test markets itself on its scientifically backed item validation, and the test delivers what the marketing promises: 30 carefully timed questions weighted across logical, verbal and spatial domains, with each item built to discriminate well between adjacent ability bands.
We recommend this test to readers who want to take their result seriously and who already understand that an online assessment is not a clinical diagnosis. The reporting is restrained and accurate, the interface free of clutter, and the score scale clearly labelled. There is no theatrical countdown or gamified animation, which is exactly what you want when you are trying to perform at your best.
Visit: reliableiqtest.com
Where most online tests return a single score, Reliable IQ Test goes further, breaking your performance down into a strengths-and-weaknesses report across four cognitive areas: verbal reasoning, numerical reasoning, spatial reasoning and short-term memory. The 30-question test takes around 25 minutes, slightly longer than average because it dedicates several items per domain to produce a reliable sub-score.
The platform credits psychology specialists with the item design, and the result is one of the more diagnostically useful online tests on this list. If you are using your IQ result to inform a career decision, a learning plan or simply your own curiosity, this format is more actionable than a single headline number.
Visit: accurateiqtest.com
Privacy on IQ-test sites is patchy. Many platforms track behavioural data well beyond what is needed to score a test. Accurate IQ Test makes a deliberate point of its privacy posture, with no requirement to register, no email capture before scoring, and minimal third-party tracking on the test pages we audited.
The test itself is a 30-question battery taking around 20 minutes, similar in structure to our overall winner but with a slightly tighter focus on logical and numerical reasoning. The scoring interface is one of the cleaner ones in the field. If discretion matters to you and you would rather not hand over personal details to take a 20-minute quiz, this is the test we recommend.
Visit: iqlevels.co
IQ Levels is the right choice if you want more than a number and a percentile. The 30-question test completes in around 25 minutes, and the report that follows runs to several pages, with breakdowns for each task type, a written interpretation of your profile, and contextual notes on what each band typically means in everyday life.
The deeper reporting comes at the cost of a slightly slower pace through the test itself, since the platform asks a short follow-up after most question clusters to build context for the final report. For users who treat their result as a starting point for reflection rather than an end in itself, this is the most thoughtful interface on this list.
Visit: officialiqtesting.com
If you would like to know how you are doing as you go, Official IQ Testing is unusual in offering periodic feedback through the test. The 25-question format takes around 15 minutes and is built around logical reasoning, problem solving and verbal comprehension. At natural pause points, the interface gives you a sense of pace and difficulty progression.
This live feedback is divisive. Some users find it motivating; others find it distracting. We mention it specifically because no other platform on this list does the same thing, and because for users new to timed cognitive testing the feedback can reduce the anxiety that artificially depresses scores.
Visit: officialiqtest.net
If you are considering a real Mensa supervised test (covered in detail later in this guide), Official IQ Test at the .net domain is a sensible warm-up. The 30-question test takes around 20 minutes and is styled to resemble the type of items used in supervised high-IQ society assessments: dense matrix problems, verbal analogies under pressure, and numerical sequences with limited time per question.
The platform compares your score against a global percentile, which is useful for setting expectations before any supervised testing. Treat this test as practice rather than as a Mensa-qualifying result. Only a supervised test in person or in a Mensa-approved remote setting can deliver a qualifying score.
Visit: officialiqtest.org
The Stanford-Binet Intelligence Scale, now in its fifth edition, is the oldest continuous IQ test in modern psychology and famous for its five cognitive factor structure. The .org Official IQ Test takes inspiration from that lineage. The 30-question test, completing in roughly 25 minutes, reports separate scores for memory, problem solving and logical reasoning rather than a single composite.
This makes it a strong choice when you want a profile rather than a number. The interface is plain, the scoring transparent, and the breakdown directly comparable to the kind of sub-domain reporting you would get from a clinically administered Stanford-Binet, although obviously without the clinician.
Visit: getiq.net
For complete newcomers to cognitive testing, Get IQ is the gentlest landing. The 25-question test takes around 15 minutes, and the platform leans on clear instructions, a friendly visual style and a short tutorial round before the timer starts. Items are pitched across the standard reasoning domains but err on the side of approachability rather than maximum difficulty.
This is not the test we would recommend for an experienced user looking to push for a ceiling score. It is, however, the test we would put in front of a curious teenager or an adult who has never sat a timed cognitive assessment and who would benefit from a low-stress first encounter with the format.
The Mensa IQ Challenge at mensa.org/mensa-iq-challenge/ is a free 30-minute practice run designed and hosted by the original high-IQ society. It is not a qualifying test, but it is an excellent benchmark against which to calibrate the scores you receive from the platforms above. If your scores across two or three of the tests on this list, plus the Mensa Challenge, consistently land at or above the 95th percentile, a supervised Mensa test in the UK is a sensible next step.
Modern intelligence testing began in Paris. In 1905, the French psychologist Alfred Binet and the physician Théodore Simon published a scale to identify schoolchildren who needed additional support, producing the first instrument that quantified reasoning ability in a standardised way. Lewis Terman of Stanford University adapted the scale for American use and published the Stanford-Binet in 1916, introducing the term "intelligence quotient" calculated as mental age divided by chronological age, multiplied by 100.
In the 1930s and 1940s a series of more sophisticated instruments appeared. The Romanian-born British psychologist John C. Raven introduced his Progressive Matrices in 1938 as a non-verbal measure of fluid reasoning, with the underlying research described in his 1936 master's thesis. David Wechsler in New York rejected the mental-age formula and built a deviation-based scale, publishing the Wechsler-Bellevue test in 1939. Raymond Cattell's Culture Fair Intelligence Test followed in 1949 in an attempt to strip cultural bias from the assessment.
The Wechsler tradition has shaped modern adult testing more than any other. The Wechsler Adult Intelligence Scale, Fourth Edition (WAIS-IV) dominated clinical practice from 2008 until late 2024, when Pearson released the WAIS-5 with normative data collected in 2023 and 2024 across the United States. The WAIS-5 is now the gold-standard adult assessment, with five primary index scores: Verbal Comprehension, Visual Spatial, Fluid Reasoning, Working Memory and Processing Speed, plus a Full Scale IQ derived from the combination.
Online IQ testing has matured alongside these clinical instruments. The best modern web-based tests do not pretend to replace a supervised WAIS-5; they offer something different, a fast, accessible, repeatable snapshot of reasoning ability for users who do not need a clinical report.
The Cattell-Horn-Carroll (CHC) theory of cognitive abilities, which now underpins almost every major IQ test, identifies several broad cognitive factors. The five most directly assessed by both clinical and online tests are below.
Fluid reasoning is your ability to solve novel problems using logic rather than learned knowledge. Pattern recognition, matrix problems and unfamiliar puzzles all draw on Gf. It is the factor most closely tied to general intelligence and the one most resistant to training.
Crystallised intelligence is the body of knowledge, vocabulary and acquired skills you carry with you. Verbal analogies and vocabulary items measure Gc, which grows steadily through adulthood and tends to be culturally influenced.
Working memory is the cognitive workbench on which you hold and manipulate information for short periods. Digit span tasks, where you repeat a sequence of numbers forwards and backwards, are the classic measure. Working memory is strongly correlated with reasoning ability and is sensitive to fatigue, anxiety and sleep deprivation.
Processing speed is how quickly you can complete simple cognitive tasks. Symbol search and coding subtests are the standard measures in clinical settings. Online tests proxy this with timed questions where speed matters as much as accuracy.
Visual-spatial processing is your ability to perceive, analyse and mentally manipulate visual information. Mental rotation tasks and block design problems both draw on this factor, which is particularly important for engineering, architecture and surgical disciplines.
IQ scores follow a normal distribution. On the Wechsler scale, the mean is 100 and the standard deviation is 15, which means roughly 68% of the population scores between 85 and 115, and roughly 95% scores between 70 and 130. The Stanford-Binet uses a similar scale with a standard deviation of 16, and the Cattell III B scale used by British Mensa uses a standard deviation of 24.
Quick answer: A "good" IQ score is anything above 100, the population mean. Scores above 115 represent the top 16% of the population; above 130 represent the top 2%; above 145 represent the top 0.1%.
The standard IQ bands are commonly described as follows:
Below 70: Very low (around the bottom 2% of the population).
70 to 84: Below average.
85 to 114: Average. This is where roughly two thirds of all people score.
115 to 129: Above average.
130 to 144: Gifted. The threshold for Mensa membership on the Wechsler SD-15 scale.
145 to 159: Highly gifted.
160 and above: Exceptionally gifted, the top 0.003% of the population.
Two important caveats. First, every score carries a standard error of measurement, usually around three to five points, which means a single score should be treated as an estimate within a range rather than a fixed figure. Second, the same person tested on different scales may receive different headline numbers depending on which standard deviation the test uses. A 132 on the Stanford-Binet (SD-16) is equivalent to a 130 on the Wechsler (SD-15), and to a 148 on the Cattell III B (SD-24). All three indicate the 98th percentile.
Online tests and clinical assessments serve different purposes and should not be treated as interchangeable. The table below summarises the practical differences.
Other clinical instruments include the Stanford-Binet Intelligence Scales, Fifth Edition (SB5), which takes 45 to 60 minutes for adults, and the WISC-V for children. A full neuropsychological diagnostic battery can run up to three hours. Online tests cannot replace these, but they are far better at one thing: giving you a low-friction, repeatable read on your reasoning ability without the cost or scheduling overhead of a clinic visit.
Not every site offering an "IQ test" deserves the name. Use this checklist before trusting a result.
No time limit on questions. Cognitive testing without time pressure does not produce a valid IQ score, because the construct of intelligence as measured by mainstream instruments specifically involves performance under time constraint.
Score pay-walled with no preview. Reputable platforms either show your score for free or display a preview before asking for payment. Tests that demand £20 to see any score at all should be avoided.
Email required before the first question. Email gates are a marketing signal, not a scientific one. The best tests on this list let you start immediately.
Promise of a specific score in under five minutes. Reliable measurement of general intelligence requires a sample of items across several reasoning domains. A five-question quiz cannot deliver this, regardless of what its headline says.
No information about the scoring scale. Trustworthy tests state whether they use a Wechsler-style SD-15 scale, a Stanford-Binet SD-16, or a Cattell SD-24. Tests that report a score with no scale information are not measuring on a meaningful axis.
Outlandish high-IQ claims. Online tests that routinely return scores above 150 to the average user are calibrated to flatter, not to measure.
For most of the twentieth century, average IQ scores in industrialised countries rose by roughly three points per decade, a phenomenon named the Flynn effect after the political scientist James Flynn, who first documented the pattern systematically in the 1980s. Recent research suggests the trend may be slowing, stalling or even reversing in some populations.
Dworak, Revelle and Condon, writing in Intelligence in March 2023, analysed responses from 394,378 Americans collected through the SAPA Project between 2006 and 2018 and found declines in verbal reasoning, matrix reasoning and letter-number series during that window. As lead author Elizabeth Dworak, a research assistant professor at Northwestern University's Feinberg School of Medicine, told ScienceDaily, the difference does not necessarily mean people are getting less intelligent: "It doesn't mean their mental ability is lower or higher; it's just a difference in scores that are favouring older or newer samples."
Independent evidence has emerged from other countries. Oberleiter and colleagues documented similar patterns in Austrian samples between 2005 and 2018, alongside a decline in the strength of the general intelligence factor itself, a phenomenon researchers call decreasing positive manifold. Most strikingly, Winter, Trudel and Kaufman published an analysis in the Journal of Intelligence in November 2024 comparing WAIS-IV (2008) and WAIS-5 (2024) normative samples and reported "a reduction in the Flynn effect from the expected value of 3 IQ points to 1.2 points" between those two normative cycles.
The takeaway for readers in 2026 is straightforward. Score increases of the kind documented for most of the twentieth century cannot be assumed to continue. This is one reason that current, recently normed assessments (whether the supervised WAIS-5 or an actively maintained online test) are preferable to older instruments using decades-old reference data.
A small set of practical decisions can swing a score by five to ten points in either direction. Before taking any test on this list, do the following.
Sleep properly the night before. Sleep deprivation has a particularly heavy effect on working memory and processing speed, two of the five Wechsler indices.
Test in a quiet, well-lit room. Background noise and visual clutter reduce attention and inflate error rates on timed items.
Use a desktop or laptop, not a phone. Spatial and matrix items, in particular, render poorly on small screens and waste cognitive capacity on visual decoding.
Complete the test in one sitting. Pausing breaks the continuity of timed measurement and almost always reduces accuracy.
Skip caffeine spikes. A normal cup of coffee is fine. A double espresso 15 minutes before a timed test usually hurts more than it helps because it inflates anxiety alongside arousal.
Do not retest within six to twelve months. Practice effects can artificially boost scores by five to ten points on a repeat administration, particularly when the item set overlaps with the first attempt.
If you take the same test twice in close succession and the scores differ by more than ten points, the more conservative figure is usually closer to your true ability. A reasonable practical strategy is to take two or three of the tests on this list within a single week and to take the median score as your best estimate.
British Mensa accepts two routes to membership. The first is a supervised in-person test, which lasts roughly two and a quarter hours and is administered at test centres across the United Kingdom. The second, introduced in recent years, is a new supervised online test that runs to roughly an hour and a half and can be taken from home with appropriate monitoring. Both are priced at £49 per the live pricing displayed at mensa.org.uk/test-your-iq/ at the time this guide was last updated.
A separate, lower-stakes option is the Mensa Home Test at £15, which gives an indicative score but is not itself qualifying. Treat the Home Test as a calibration exercise.
To join British Mensa, you must score at or above the 98th percentile on a Mensa-recognised test. On the Cattell III B scale used in supervised UK testing, the qualifying score is 148, which corresponds to roughly 132 on the Stanford-Binet SD-16 and 130 on the Wechsler SD-15. All three figures represent the same underlying percentile.
If you scored consistently in the top 2% on multiple online tests in this guide, plus the free Mensa IQ Challenge linked above, a supervised UK Mensa test is the logical next step.
What is the most accurate IQ test in 2026? The most accurate IQ test available remains a supervised clinical assessment using the WAIS-5 (for adults) or the WISC-V (for children), both published by Pearson. Among the online tests in this guide, Official IQ Tests is our top pick for overall accuracy, with IQ Certificate and Real IQ Test close behind.
Are free online IQ tests reliable? Well-built free tests are reliable enough for personal interest and rough self-assessment, with test-retest correlations typically between 0.70 and 0.85. They are not a substitute for supervised clinical testing if you need a diagnostic or legal-grade result.
How long does an online IQ test take? Most online tests on this list complete in 15 to 30 minutes. Clinical assessments are considerably longer: the WAIS-5 runs 60 to 90 minutes, the Stanford-Binet Fifth Edition runs 45 to 60 minutes, and a full neuropsychological battery can run up to three hours.
What is an average IQ score? The population mean on the Wechsler scale is 100, with a standard deviation of 15. Roughly 68% of people score between 85 and 115, and roughly 95% score between 70 and 130.
What IQ qualifies for Mensa in the UK? British Mensa requires a score at or above the 98th percentile. On the Cattell III B scale used in supervised UK testing, the qualifying score is 148. The equivalent thresholds are 132 on the Stanford-Binet SD-16 scale and 130 on the Wechsler SD-15 scale.
What is the difference between an online IQ test and a clinical WAIS-5 assessment? Online tests are self-administered, take 10 to 30 minutes, cost between nothing and roughly £40, and produce a single composite score with a brief report. A clinical WAIS-5 is supervised by a registered psychologist, takes 60 to 90 minutes, costs £150 to £1,500 or more privately in the UK, includes hands-on tasks and produces a detailed written report covering five cognitive indices.
Can I take a real IQ test from home? You can take a well-normed online IQ test from home with no supervision. For a qualifying Mensa score, British Mensa now offers a supervised remote test alongside its traditional in-person option, both priced at £49 as of May 2026.
Do online IQ tests give certificates? Several do. Every test on this list offers some form of result documentation, and IQ Certificate in particular produces a downloadable, dated certificate suitable for personal records and high-IQ society practice.
Can you increase your IQ? Targeted training tends to improve performance on the specific tasks you practise without lifting general intelligence as measured by independent tests. Sustained changes to sleep, exercise, nutrition and cognitive engagement do appear to influence measured scores at the margin, particularly on tasks that involve working memory and processing speed.
Is the Flynn effect still happening? Possibly not in the way it used to. Recent studies in the United States and Austria show stagnating or declining scores on several reasoning subtests since the mid-2000s, and the WAIS-5 normative cycle showed only a 1.2-point Flynn effect compared to the expected three points based on prior decades.
How often should I retake an IQ test? No more than once every six to twelve months on the same instrument. Practice effects can artificially inflate a repeat score by five to ten points if the gap is too short.
Are online IQ tests culture-fair? The non-verbal tests on this list (Actual IQ Test, Real IQ Test) approach culture-fair status by relying heavily on matrix and pattern items. Tests with strong verbal components inevitably reflect English-language familiarity to some degree.
Which IQ test is best for children? None of the tests on this list is normed specifically for children. For children, the clinical WISC-V remains the gold standard. If you are looking for a child-suitable cognitive practice tool, look for a platform with explicit age-band norming rather than adapting an adult test.
How much does a professional IQ test cost in the UK? A privately administered WAIS-5 typically costs between £150 and £1,500 depending on the clinician, the report length and the region. NHS cognitive assessments are available in clinically indicated cases (for example, in the context of a learning disability or dementia assessment) but are not offered for general curiosity.
If you take only one test from this list, take Official IQ Tests. It is the strongest all-round option, with a properly varied 30-item battery, transparent scoring and a clean post-test report.
If you want a credential you can print and share, choose IQ Certificate. If English is not your first language, or if you suspect verbal items have unfairly weighted your previous scores, Actual IQ Test is the culture-fair choice. For a deeper diagnostic-style breakdown across cognitive sub-domains, Reliable IQ Test and IQ Levels deliver more interpretation than any other platform on this list.
For users planning to attempt a real Mensa assessment, treat Official IQ Test (Mensa-styled) and the free Mensa IQ Challenge as practice, then book the supervised Mensa test at mensa.org.uk/test-your-iq/ when you are consistently scoring in the top 2%.
And for everyone, the same two reminders apply. First, no single online score is your "real" IQ; take two or three tests and look at the median. Second, an IQ score is a description of one slice of cognitive ability at one moment, not a verdict on you. Treat the result as useful information, not as identity.
This guide was last reviewed and updated on 20 May 2026. References used: APA Standards for Educational and Psychological Testing (2014); Pearson WAIS-5 Technical and Interpretive Manual (Wechsler et al., 2024); Mensa International and British Mensa public information; Dworak, Revelle and Condon, Intelligence (2023); Winter, Trudel and Kaufman, Journal of Intelligence (2024). All outbound links to testing platforms were verified live in May 2026.