The Mensa IQ Challenge score, and what it can tell you

0 The short answer

The Mensa IQ Challenge is a free practice test on mensa.org with 35 puzzles to solve in 25 minutes, and Mensa states on the page that it is provided for entertainment purposes only and that your score will not qualify you for Mensa. The page publishes no norms, no reliability and no item format, and it shows no number at all if you score under 100, so the result is a rough indication from one short, unsupervised session and not an IQ you can quote. A score that counts for Mensa needs an approved test that was properly administered and supervised, for example American Mensa local group testing at 60 dollars or private testing at 99 dollars (both read on November 4, 2026). ACIS Quick is one adult normed option at 25 dollars and about 45 minutes (read on November 4, 2026, and prices can change), and the decisive limit is that no online test, the Challenge and ACIS included, can substitute for the supervised sitting.

1 What Is the Mensa IQ Challenge, and What Does Its Own Page Say?

The Mensa IQ Challenge is a free online test hosted by Mensa International whose own page describes it in three sentences of substance: 35 puzzles, 25 minutes, and difficulty that rises as you go. We opened the Mensa IQ Challenge page on November 4, 2026. It states: "This challenge involves trying to solve 35 puzzles within 25 minutes." It adds that the puzzles get progressively difficult, and it warns that this particular challenge is "actually pretty tough". It says you can take the practice test as many times as you want, and that the engine supports multiple languages selected from your browser preferences: English, Chinese (Traditional and Simplified), French, German, Hungarian, Icelandic, Italian, Macedonian, Norwegian, Polish and Spanish.

The page also credits the people behind it. It names Olav Hoel Dørum as the author of the test questions and scoring and Tedd Hansen as the developer of the test engine and web pages, and it carries a Mensa International Limited copyright line for 2026. That matters for one reason: the page presents the Challenge as a practice test with named authors that Mensa hosts, not as a Mensa admission instrument, and it calls itself entertainment in its own words.

What the page does not say is as informative as what it says. It does not state the type of puzzle (visual, verbal or numerical), the age bands, the upper limit of the score, the population the score is compared with, the year the comparison group was tested or any reliability figure. We asked the page directly about each of these and record every one as not stated. Nothing on it tells a taker how precise the number is, and without that, a score of 120 and a score of 126 cannot be told apart in any way the page supports.

The surrounding Mensa pages fill in the purpose. The FAQ on getting your IQ tested recommends the Challenge to people whose national Mensa does not offer a pre-test, as an online practice option, and says taking a practice test can give "a boost of confidence before the real exam". That is the job the test was built for: a free rehearsal that tells you whether the format suits you, not a measurement of your intelligence. If you want the wider map of free options and how they differ, the guide to free IQ tests sets them side by side, and the page on what IQ test questions look like explains the item families without reproducing any item.

One caution about the 35 and the 25. The same two numbers appear on the Mensa Norway online test, which is a separate page with its own wording and its own score range. The comparison table further down sets them side by side. Here the point is narrower: a 35 puzzle test in 25 minutes is a short test, and short tests carry larger errors of measurement, a point the Standards for Educational and Psychological Testing make about any shortened assessment.

2 Is the Mensa IQ Challenge an Official Test, and Can It Qualify You for Mensa?

No: Mensa says in three places on its own site that the Challenge is practice, and it says that online tests cannot be used for admission. On the Challenge page itself the sentence reads: "This Mensa IQ Challenge is provided for entertainment purposes only, and your score will not qualify you for Mensa." The same page then suggests that a taker who is pleased with the score consider taking a "properly administered and supervised IQ test". On the Mensa FAQ about getting your IQ tested (read on November 4, 2026), the answer to "How do I qualify for Mensa?" includes: "Please be aware that online tests cannot be used for admission to Mensa." The FAQ also states that a pre-test is just for practice and cannot be used to qualify "even if you score at or above the 98th percentile", and the section recommending the Challenge repeats that it is practice only.

What does count? The same FAQ says membership is open to people who have attained a score within the upper two percent of the general population on an approved intelligence test that has been properly administered and supervised. You can either take a test administered by Mensa or submit a qualifying score from another approved test, and the FAQ adds that Mensa only accepts tests given by people qualified to carry out testing privately. The price and the format are set country by country: the FAQ says the price of Mensa tests varies from country to country, and the Mensa national groups page lists the national groups and tells residents of countries without one to contact Mensa International. Rules differ by country, so every national fact below is labeled with its country.

American Mensa is a good example of what a supervised route costs. Its testing page, read on November 4, 2026, lists Local Group Testing at 60 dollars and Private Testing at 99 dollars (regular prices; the page also says promotional discounts appear at checkout, and we do not print promotions). It describes the Private option as administered electronically at approved testing facilities, and states that the exam takes between one and two hours. It defines a qualifying score as one that indicates you have tested at or above 98 percent of the general population. For the Mensa routes in four countries, with fees and dues, see how to join Mensa, and for what the threshold means in IQ points see the page on the score Mensa needs.

Mensa's own page on what IQ is explains the numeric side. It says Mensa admits the top two percent and accepts many tests, "as long as they have been standardised and normed", and it gives two examples of how the same percentile lands on different numbers: 132 or more on the Stanford-Binet and 148 or more on the Cattell equivalent. The page does not say why the two cutoffs differ, but they are what you would expect from tests that use different scales. On a scale with a standard deviation of 15, the top two percent starts near 131 (our arithmetic, from the normal curve), and the explanation of why SD 15 matters shows how the conversion works. The consequence for this page is direct: a number from the Challenge cannot be compared with any Mensa cutoff unless the Challenge says which scale it uses, and its page does not.

The FAQ also states that Mensa International is finishing preparations for computerised adaptive testing, and that the adaptive test will not have a fixed set of items. That is a statement about a future admission format, not about the Challenge, and we do not read anything into it for the Challenge beyond the obvious: the practice test and the admission instrument are different things. Siblings of this page cover the country routes in detail: the Mensa home test page covers the pre-tests some national groups send out, and the pages on the Mensa Denmark test and the Mensa Germany test cover those two national groups.

3 How Is the Challenge Scored, and Why Does It Show No Number Below 100?

The page does not describe how the score is calculated, and it states only one rule about the result: if you do not get 100 or more, the automated answer says that your score lies outside the range the test can currently measure. The page says that at the end of the test, if you do not get a score of 100 or more, the automated answer will say: "Your score lies outside the range that the test is currently able to measure." So the Challenge has a display floor at 100. Anyone who performs below the level it maps to 100 receives a message and no number.

A display floor is a design choice with consequences for reading the result. First, it removes information in the half of the range where half of all adults sit: on a scale where 100 is the average, roughly half of people would score under 100, and for them the test returns nothing quantitative. Second, it makes the numbers that are shown harder to interpret, because the reader never sees how the test treats the lower half of its own scale. Third, it means that a forum thread of Challenge results is a biased sample by construction: people who were shown a number have something to post, and people who were shown the message have no number to share. That is a statement about how the display works, not about any particular thread, but it is why a tally of posted scores says little about how the test behaves.

The page says nothing about a ceiling. Whether the test can separate a 130 from a 145, or reports anything above a point, is not stated. The Norway test, by contrast, states a range of 85 to 145, which at least tells a taker where its scale ends. A test with no stated ceiling and no published reliability leaves a high scorer unable to tell how far above the average a displayed number really sits. The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014), which we read in the open access edition, ask that the test developer set forth how scores are intended to be interpreted (Standard 1.1) and that those responsible for releasing scores describe in simple language what the test covers, what the scores represent, the precision or reliability of the scores and how they are intended to be used (Standard 6.10). The Challenge page partly does the first of these by calling itself entertainment, and leaves the rest unstated.

The page also contains one idea that every taker should keep: it is common to get different scores on IQ tests, even when it is the same test taken at different times. That is Mensa's own acknowledgment of measurement error, and it is the right frame. An honest reading of any number the Challenge shows is a band around it, not a point, and the width of that band is not published. How a published band is built for professional tests is explained in the page on reliability and validity, and how to read a score and its band together is in the guide to what IQ scores mean.

Finally, the language of the message is worth reading closely. "Outside the range that the test is currently able to measure" does not say you scored low. It says the test cannot place you. A taker who is told this has learned that the instrument is poorly matched to their performance on that day, which is different from learning anything about their intelligence. Treating it as a verdict on ability is a tempting misreading of this screen, and nothing on Mensa's page supports it.

4 How Much Can a 25 Minute, 35 Puzzle Score Tell You?

It can tell you something real about one narrow skill under time pressure, and almost nothing about the rest of what an IQ score summarizes. The reasoning has three parts: how much of the construct the test samples, how precisely, and under what conditions.

On coverage, the Challenge page does not say what the puzzles are. If they are visual pattern puzzles, as the Norway test page describes its own 35 puzzles ("in the form of visual patterns"), then the test samples fluid reasoning and in particular induction, the narrow ability to discover the rule behind a set of regularities. Even a well chosen matrix style test samples that ability and little else. The title of Gilles Gignac's 2015 Intelligence paper states the problem: Raven's is not a pure measure of general intelligence: Implications for g factor theory and the brief measurement of g. Our practical reading is that a single matrix test is a useful indicator and not a stand-in for a battery. Six broad abilities are tracked by a modern battery, and a one format test leaves vocabulary and knowledge, quantitative reasoning, visual spatial skill, working memory and processing speed unmeasured. The page on the CHC model lays out those domains.

On precision, the Standards state the general principle in their reliability chapter: if an assessment is shortened by decreasing the number of items or tasks, its reliability is likely to decrease, and if it is lengthened with comparable tasks, reliability is likely to increase. Thirty five puzzles is a short sample. This is a general psychometric statement and not a measured figure for the Challenge, whose reliability the page does not report. It is also why a professional battery spends hours, and why our own Quick form, at about 45 minutes, comes with a confidence interval rather than a bare number.

On conditions, the Standards add a comment under Standard 3.4 that applies directly: "unproctored test administrations where standardization cannot be ensured" could provide an advantage to some test takers over others, and in these situations "test results should be interpreted with caution". A Challenge sitting happens on your own device, in your own room, with a timer the page controls and nobody checking who you are, how many times you have seen the puzzles or whether you looked something up. The same comment notes that examinees on older, slower equipment may be unfairly disadvantaged. For a practice test, none of that matters. For a number you plan to quote, all of it does.

Taken together, the honest ceiling on the claim is modest. A Challenge result that shows a number suggests you handle this kind of puzzle well in this sitting. It does not tell you your Full Scale IQ, your percentile among adults, or how you would score in a supervised sitting, and the page does not claim that it does. The pages on whether online IQ tests are accurate and how to tell whether an IQ score is real give the checklist that any online score should pass, and the Challenge, by its own description, does not try to.

5 How Does the Challenge Differ From the Mensa Norway Test and a Supervised Sitting?

The Challenge and the Norway test are two separate free online pages with the same puzzle count and time limit and different wording about their scores, and both differ from a supervised sitting in the thing that matters most: someone qualified watches the session. We opened both pages on November 4, 2026 and set out what each one states.

The Norway row comes from the Mensa Norway online test page, read on November 4, 2026. It says the test consists of 35 puzzles in the form of visual patterns that must be solved within a 25 minute time limit, gives an indication of general cognitive abilities represented by a score between 85 and 145 where 100 is the population average, and is not a substitute for professional intelligence tests such as those administered by Mensa and licensed psychologists. The page for the Norway test is also where the practical difference lies: it states a score range with a floor of 85 and a ceiling of 145, while the Challenge states a floor at 100 and says nothing about a ceiling. The full account of the Norway test, including how accurate its score is, is on the page about the Mensa Norway IQ test.

Do the two share items or norms? The pages do not say. They share the puzzle count, the time limit and a creator credit (the Challenge page names Olav Hoel Dørum for the questions and scoring), and these are suggestive. But "suggestive" is not "stated", and a reader cannot assume that a score on one maps to the other. If you took both and the numbers differed, the Challenge's own sentence about different scores at different times is the right default explanation, together with ordinary measurement error and the retest effect discussed below.

The third column is the one that matters for qualifying. Mensa's FAQ describes the supervised route as an approved test, properly administered and supervised, given by people qualified to carry out testing, with the price and the format varying by country. The supervised sitting differs from both online tests in who checks the identity of the taker, the equipment, the instructions and the timing, and in whether the result is compared with a described norm group. The page on the pre-tests that some national groups send home explains where a take home practice sits between the two.

6 Why There Is No Answer Key for the Mensa IQ Challenge on This Page

No answers are given here, because an answer key for a practice test that you have not taken is worth nothing, and one you have used before taking it destroys what the score was for. People type "mensa iq challenge answers" for three different reasons, and each has an honest reply.

If you are stuck on a puzzle mid test, the Challenge is built to be hard, and the page warns that it is "actually pretty tough". Getting stuck is the test working as designed, since the puzzles get progressively more difficult. If you finished and want to check your work, the page gives the automated result at the end, and you may retake it as often as you like, so a second attempt gives you the feedback without an answer list. If you want to see the answers before taking it, the score you receive afterwards is not a measurement of anything, because the test measures how you solve unfamiliar problems and you will have solved nothing.

There is also a quality problem with answer lists. Whether the Challenge serves the same puzzles each time, a rotating set or different items per language is not stated on its page, so a list written for one version may not match another. A page claiming to hold the answers cannot be checked against the live test by a reader who has not already taken it. And there is a security reason that the Standards address: under Standard 3.4, "unauthorized distribution of items to some examinees but not others" can threaten the validity and fairness of score interpretations. Mensa's FAQ gives the same reason for its direction of travel, saying older paper based tests may be nearing the end of their useful life partly because of age and the risk of items leaking online, and that its adaptive test will not have a fixed set of items. We do not reproduce items, sample questions or answers from any real test and we do not link to sites that do.

Practice has a measured effect that is worth knowing about. A meta-analysis by Scharfen, Peters and Holling in Intelligence found significant retest effects and no further score gains after the third test administration, from 174 samples in 122 studies with a total of 153,185 participants; the size of the effect depended on the cognitive ability tested, the equivalence of the test forms, the retest interval and participant age (details in the abstract of Scharfen and colleagues). The Challenge allows unlimited attempts, so a taker who retakes it is partly measuring practice. If you want to understand how much a second score typically rises and why, the page on retaking an IQ test covers it with the evidence.

What you can do instead, if the aim is a score you can trust, is spend the time on a test that was normed against adults and that states its precision. If the aim is only to see whether you enjoy this style of puzzle, take the Challenge once, without help, and treat the number, if one appears, as a game result.

7 What Are the Real Options for Taking an IQ Test, and What Does Each Give You?

The real options split by who supervises and who the score is compared with, and the Challenge sits in the free, unsupervised, norm unstated corner. The table sets out each option with what we could read about it. Prices are regular prices read on November 4, 2026, and ACIS prices can change too.

Two rows need a comment. The psychologist row is the benchmark for a clinical or legal use, but we did not read a price for it today, so the table says so instead of guessing; the guide to every option for taking an IQ test compares costs and settings in full. The game style row is deliberately general: we are not describing any named product, only noting that screeners that do not publish a norm group cannot tell you what a number is a number of.

The ACIS rows are our own, and they are limited by design. All three forms are online and unsupervised, and ACIS states its limits on its research page. The forms differ in depth, not in status. The page on the best online IQ tests compares the field for adults, and the guide to IQ tests for adults explains how to pick among them. If your concern is the high end, the page on high range IQ tests explains what any standard test can and cannot separate near the top.

The choice among the first and last rows is a question of purpose. If the goal is membership, the supervised row is the only one that counts. If the goal is to learn whether you are likely to be near a society's threshold before you pay its fee, an adult normed battery gives a more usable answer than a 25 minute puzzle, with the caveat that no online result predicts what a supervised sitting will return. If the goal is curiosity, any of the free options is fine, and the Challenge is a reasonable one.

8 What Does the Challenge Measure, and What Does It Leave Out?

If the Challenge is a pattern puzzle test, it samples fluid reasoning, Gf in the Cattell Horn Carroll model, and specifically the narrow ability of induction, and it leaves out the five other broad domains that a battery covers. The caution in that sentence is deliberate, because the Challenge page does not state its puzzle type; we are drawing the inference from the Norway page's description of a sibling test and from the Challenge's own framing as a puzzle test with rising difficulty. Treat it as likely and unconfirmed.

In the CHC model, which the page on the CHC model explains in full, fluid reasoning (Gf) is the ability to solve unfamiliar problems by detecting relationships and applying rules rather than retrieving learned facts. Induction is the narrow ability to observe a set of instances and infer the underlying rule. A timed matrix style task measures that ability while also drawing on visual perception, working memory to hold candidate rules, and speed. That is why a score from such a task is only partly Gf and partly something else; the explanation of what a fluid intelligence test measures separates those parts, and the overview of what IQ measures shows where fluid reasoning sits within g.

What the format leaves out is large. Gc, crystallized knowledge and verbal comprehension, is excluded on purpose by a visual puzzle. Gq, quantitative reasoning, is not sampled. Gv, visual spatial processing beyond pattern completion, is barely sampled. Gwm, working memory capacity, and Gs, processing speed, appear only as side demands. The guide to cognitive domains describes all six, and the difference between them is not a technicality: the page on index score differences explains how to read gaps between a person's indices, which is why one domain can say little about another in the same person.

ACIS offers two subtests in the same ability family, and describing them is useful because they show what a fluid reasoning measure looks like when it is one part of a battery. On the Matrix Reasoning subtest page, ACIS describes matrix tasks as belonging to the tradition of nonverbal abstract reasoning, where a person studies a pattern of shapes, notices how rows or columns change and infers the missing element, and it places the task under fluid reasoning (Gf) in CHC terms. Matrix Reasoning and Figure Weights are two of the five subtests that make up the ACIS fluid reasoning index. Both are in the Quick form, next to Similarities, Vocabulary, Digit Span and Alphanumeric Sequencing, so a Quick result samples reasoning, language and working memory instead of one puzzle type.

9 How Should You Read a Score From a Short Unsupervised Test?

Read it as a band, not a point; read it against a described norm group, not a headline; and read it as dated, because norms age. Each of those three reads has a source behind it, and each applies to the Challenge number and to any other online score.

A band, not a point. Every score has an error of measurement. A reported 95 percent confidence interval is the standard way to show it, and the Standards (Standard 6.10, comment) note that score precision might be depicted by error bands or likely score ranges showing the standard error of measurement. A test that publishes no interval leaves the reader to guess how wide the band is, and a short test usually has a wider band than a long one. The page on how IQ percentiles relate to scores explains why a point and a rank are not the same thing, and the IQ percentile calculator turns a score into a percentile under the SD 15 convention.

A described norm group. A norm referenced score means nothing without the group it is referenced to. The Mensa page on IQ describes the logic: the peer group score is an IQ of 100, obtained by applying the test to huge numbers of people. The Challenge page names no such group. Scores on a scale with a mean of 100 and a standard deviation of 15 correspond to fixed percentiles; for example, a score of 130 is 2 standard deviations above the mean, which is about the 97.7th percentile of the normal distribution (our arithmetic). That conversion holds only when the test is actually scaled that way, which the Challenge does not say. How norms are built for professional tests is covered in the page on how IQ scores are normed, and the rarity calculator shows how rare a given IQ is on that convention.

A dated number. Norms drift. Pietschnig and Voracek's meta-analysis of the Flynn effect, One Century of Global IQ Gains in Perspectives on Psychological Science, covered 271 independent samples from 31 countries and estimated gains of 0.41 IQ points a year for fluid tests, with gains that vary by domain and have decreased in recent decades. Taken at face value, that is a bit over 4 points a decade on fluid measures (our arithmetic on their estimate), which is why a test scored against an old comparison group would tend to inflate scores for current takers. The Challenge page does not give the year of its comparison group, so the reader cannot judge that drift either.

There is a fourth read that the Challenge invites: the retake. The page allows repeating the test, and Scharfen and colleagues found that gains from repeated testing are real and flatten after the third administration. A first sitting is the cleaner one, and a later sitting measures practice plus ability.

10 What Practical Conditions Affect an Unsupervised Session?

Conditions that a proctor controls in a testing room become your responsibility at home, and for a 25 minute timed test they can move the number. The Standards make the point generally. Standard 3.4 and its comment ask that test takers receive comparable treatment and that those responsible adhere to standardized administration, and the comment on computerized testing says that technology itself should not influence scores, with older or slower equipment possibly disadvantaging some examinees. A home sitting cannot guarantee any of that, so the number carries the caveat, and your job is to make your own conditions as close to a testing room as you can.

For the Challenge, what you control is the 25 minutes. The page does not state whether the timer can be paused, whether the sitting survives a refresh or what happens if the connection drops, so the sensible plan is to treat the session as one uninterrupted block. Choose a time when you are alert, not at the end of a long day, since a timed test deserves your best attention. Silence notifications, close other tabs and use a screen large enough to see a whole pattern without scrolling. Use a mouse or trackpad you are used to, since a clumsy input device costs seconds in a timed test.

If you decide to take an ACIS form instead, the conditions are documented. Three subtests, Digit Span, Alphanumeric Sequencing and Arithmetic, present their items as audio and need working sound, so test your speakers or headphones first. Layer Rotation carries no words at all, as the technical manual states. You may take breaks and your progress is saved between subtests, with 30 days to complete a form, which suits the longer batteries better than a single 25 minute window. A quiet room matters for the working memory subtests in particular. See the Digit Span page to know what that subtest asks before you sit it. None of this makes an online session supervised, and ACIS does not claim that it does.

11 What Should You Do With Your Challenge Result?

What you do next depends on the screen you saw: a number, or the message that your score lies outside the range. Each branch has a sensible action, and none of them is to quote the result.

If you saw the message, you have learned that this test, on this day, could not place you. That is not a finding about your intelligence, and it is not a reason to avoid a supervised test. If you are curious, take the Challenge again after a break, or try an instrument that reports a score across the whole range with an interval. The page on the quick IQ test explains what a short normed battery costs in time and precision, and the free trial lets you try parts of ACIS with no card.

If you saw a number you like, remember what it is: one reading from a 35 puzzle practice test with no stated norm group. Mensa's own suggestion is to take a properly administered and supervised test if you are pleased with your score. Before you do, find your national group on the Mensa national groups page, read what its test costs and how it is run, and check whether it offers a pre-test. Fees, formats and accepted prior tests differ by country; American Mensa's page, for example, invites people who have taken all available tests without qualifying to submit evidence of prior testing, and it does not list which prior scores it accepts, so ask the group before assuming any score you already hold counts. The page on how to join Mensa walks through the routes and the dues in four countries.

If you are not sure you want to pay a society's fee, you can get an adult normed picture first. That picture is not a prediction. It tells you where you stand by index relative to adults of your age, and a profile that shows strong fluid reasoning but average processing speed, for example, tells you more about how you think than a single number does. It is a reasonable basis for deciding whether the fee is worth your money, and for nothing else. The page on gifted tests for adults explains how adults use a normed result to decide on that next step.

If you took the Challenge for fun, nothing more is needed. Mensa itself presents it as entertainment, and a free puzzle you enjoyed is a fine outcome. Share the number if you like, but not as an IQ, and do not write it on an application or a CV. The Standards (Standard 9.4) put the responsibility on the user: when a test is used for a purpose for which little or no validity evidence is available, the user is responsible for documenting the rationale and obtaining evidence of reliability and validity for that use. A practice test with no published evidence cannot meet that bar for a membership claim, a hiring claim or a public claim about your intelligence.

12 Where ACIS Sits, and What a Buyer Should Do

ACIS is an online, unsupervised, adult normed battery that gives a Full Scale IQ and six indices, and it is not a route into Mensa, a clinical instrument or a predictor of a supervised result. We say that once and plainly: ACIS is not accepted by any high IQ society and does not predict a pass, and nothing here suggests that Mensa runs, endorses or accepts it. The research page states the limit in the site's own words: ACIS "is not intended for clinical diagnosis, hiring decisions, accommodation determinations or admission to high IQ societies". It is online and unsupervised, offered in English only, and its manual notes that verbal and knowledge tasks are affected by English proficiency and educational exposure. For documentation, see the technical manual and the ACIS page for research use.

What ACIS offers this reader is narrow and concrete. It reports a Full Scale IQ with its 95 percent confidence interval and six index scores on the standard scale (mean 100, standard deviation 15), with scaled subtest scores from 1 to 19, percentiles and adult norms for ages 16 to 90. The Quick form costs 25 dollars and takes about 45 minutes, Optimized costs 35 dollars and about 110 minutes, and the Full Scale form costs 59 dollars and about 175 minutes, all one time with no subscription (prices read on November 4, 2026; they can change). You can take breaks, progress is saved and you have 30 days to finish. The free trial needs no card, and one is allowed per person. The refund promise is conditional and should be read as written: "Full refund within 5 days if the report does not deliver the features described at checkout or a technical issue prevents you from accessing your results."

How should a score from any online test be read? The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014) ask that the intended interpretation of scores be stated clearly with evidence (Standard 1.0), warn that unproctored administration calls for cautious interpretation (comment on Standard 3.4) and ask that the precision of scores be described to the people who receive them (Standard 6.10). The American Psychological Association's Guidelines for Psychological Assessment and Evaluation (2020) add that a psychologist should recognize when the selection and use of a test deviates from its expected or intended purpose and weigh the implications when interpreting results. Using an online battery to decide whether to book a supervised sitting is within the purpose we describe; using any online result as proof of eligibility is outside it.

The next step is practical. If you only wanted the puzzle, you are done. If you want a score that reports its own precision, start the free trial and open an example report before you pay for a form. If you intend to apply to Mensa, read your national group's page, book its test and treat an ACIS result as context only, not as a substitute for the supervised sitting that Mensa requires.

13 Sources Behind This Page

Every third party fact and figure on this page comes from the documents below, opened or verified on November 4, 2026, and any figure we calculated ourselves is labeled in the text as our arithmetic. The Mensa pages were read as published that day, prices and policies change, and where a document could not be fetched directly the entry says how it was read.

1. Mensa International. Mensa IQ Challenge (mensa.org page). Read November 4, 2026. https://www.mensa.org/mensa-iq-challenge/
2. Mensa International. Getting your IQ tested: FAQs (mensa.org page). Read November 4, 2026. https://www.mensa.org/getting-your-iq-tested-faqs/
3. Mensa International. What is IQ? (mensa.org page). Read November 4, 2026. https://www.mensa.org/what-is-iq/
4. Mensa International. National groups (mensa.org page). Read November 4, 2026. https://www.mensa.org/national-groups/
5. Mensa Norway. Online test (Mensa Norway test page). Read November 4, 2026. https://test.mensa.no/
6. American Mensa. Testing (us.mensa.org page, regular prices). Read November 4, 2026. https://www.us.mensa.org/join/testing/
7. American Educational Research Association, American Psychological Association and National Council on Measurement in Education (2014). Standards for Educational and Psychological Testing. Washington, DC: AERA. Open access edition. https://www.testingstandards.net/open-access-files.html
8. American Psychological Association (2020). APA Guidelines for Psychological Assessment and Evaluation. APA Task Force on Psychological Assessment and Evaluation Guidelines, approved by the APA Council of Representatives, March 2020. https://www.apa.org/about/policy/guidelines-psychological-assessment-evaluation.pdf
9. Scharfen, J., Peters, J. M., and Holling, H. (2018). Retest effects in cognitive ability tests: A meta-analysis. Intelligence, 67, 44 to 66. https://doi.org/10.1016/j.intell.2018.01.003
10. Pietschnig, J., and Voracek, M. (2015). One century of global IQ gains: A formal meta-analysis of the Flynn effect (1909 to 2013). Perspectives on Psychological Science, 10(3), 282 to 306. https://doi.org/10.1177/1745691615577701
11. Gignac, G. E. (2015). Raven's is not a pure measure of general intelligence: Implications for g factor theory and the brief measurement of g. Intelligence, 52, 71 to 79. https://doi.org/10.1016/j.intell.2015.07.006

14 Frequently Asked Questions