How to tell a legit IQ test online before you pay for one

0 The short answer

An online IQ test is legit to the degree that you can verify it: who built it, which adults your score is compared with, whether a technical manual exists, what it costs and what the refund says, all before the first item. Those checks measure transparency and not quality, because a page can document a weak test and a good test can sit behind a thin page. ACIS Full Scale costs 59 dollars for about 175 minutes, Quick 25 dollars for about 45 minutes and Optimized 35 dollars for about 110 minutes (prices read on November 4, 2026), and the section on ACIS shows what it publishes on each of the ten checks. The decisive limit is that any unsupervised online score, ACIS included, is a personal estimate and not a credential for a clinic, employer, school or society.

1 What Does "Legit IQ Test" Actually Mean?

A legit IQ test is not a test with a badge, because a badge proves nothing you cannot inspect: it is a test whose builder, norms, evidence, price and limits you can inspect, and the word bundles three separate questions that need three separate checks. The first question is about the seller: is someone accountable by name, and is the money side honest and visible. The second is about the test: is it documented well enough that the number it returns can be interpreted. The third is about you: is the score fit for what you want to do with it. A site can pass the first and fail the second, and a well documented test can still be the wrong tool for the third.

A stamp of approval is not the benchmark. The professional benchmark is the Standards for Educational and Psychological Testing, the joint document of the American Educational Research Association, the American Psychological Association and the National Council on Measurement in Education. Its 2014 edition, opened for this page on November 4, 2026, states in its legal disclaimer that it is not enforceable by the sponsoring organizations, while noting that regulators and courts have recognized it as setting out the generally accepted professional standards that developers and users of tests follow. That is the right way to hold it: a yardstick any reader can apply, with no one at the door checking who did.

The word also gets pulled in a commercial direction, because a page that judges whether a test is legit can have a stake in the answer. This page has that structural problem. It is published by ACIS, which sells an IQ test, so the ten checks below are written to be run on ACIS as well, and the section on ACIS shows what it publishes on each one.

Four neighboring questions are answered elsewhere, and this page does not repeat them. How accurate a given score is belongs to the page on how to tell whether an IQ score is accurate. What the research on online administration shows belongs to the page on whether online IQ tests are accurate. How billing models work, trial to subscription to one time payment, belongs to the page on paid IQ tests. And the comparison of free tools with validated ones belongs to the page on free against validated tests. What stays here is the narrower job: judging the seller and the test before you pay.

2 How Do the Real Options Compare Before You Look at Any Site?

The options differ on five things that matter more than brand: who the score is compared with, whether a person supervises you, how long it takes, what it costs and what you actually receive. The table sets out the main ones with prices and descriptions read on the publishers' own pages on November 4, 2026. Prices change, ACIS prices included.

Three details in that table deserve a closer look. The first is that the supervised routes are not the same product as an online battery. Pearson describes the WAIS-5 on its publisher page as an individually administered clinical instrument for assessing cognitive ability, with ages 16:0 to 90:11 and a qualification level of C. The clinic fee comes from the University of South Florida clinic fee page, which lists gifted assessment with the WISC or WAIS, with no edition named, at 100 to 400 dollars on a sliding scale with half due at the first meeting. That is one clinic, and the page on official IQ test routes covers the others.

The second is that the Mensa row answers a different question from the one this page asks. Mensa's testing page says its exam is given for admission into Mensa and not to quantify intelligence. It is a legitimate route to a membership decision and it is not a route to an IQ number. The third is that a research resource can be fully documented and still unusable for your purpose. The ICAR project page says, in answer to whether it can determine an individual's IQ, that you should not, and that without norms you cannot infer an individual's IQ. Documentation and usability for you are different properties, which is why the checks below are ten and not one.

For a wider map of where to take a test, the page on where to take an IQ test lists every route, and the guide to the best online IQ tests compares the online options without assigning numerical rankings. The sibling page on which IQ test to take starts from your purpose rather than from the seller.

3 What Are the Ten Checks, and What Can Each One Prove?

Each check reads one page on a site, takes a minute or two, and proves something narrow: that a piece of information exists, never that the test is good. The honest way to use the list is as a set of filters that move your confidence up or down, not as a score. The table gives all ten at a glance; the next four sections work through them.

Checks 1 to 4 protect your interpretation: they tell you whether the number can be read at all. Checks 5 to 7 protect your money. Checks 8 to 10 protect your expectations. A site that fails 5 to 7 and passes 1 to 4 has a documented test sold badly, and a site that passes 5 to 7 and fails 1 to 4 has a transparent store with a number you cannot read. Neither pattern is a verdict on its own.

The page on the scientific IQ test goes deeper on the research side with a longer list of checks, and this page is the faster version meant for the moment just before a payment screen.

4 Checks 1 and 2: Who Signed the Test, and Whom Does It Compare You With?

The first two checks are the cheapest and the most revealing: a test you cannot trace to accountable people, or whose score has no described reference group, cannot be interpreted however polished it looks. Open the About page, the footer and the page that explains the score, and look for two things: names, and a sentence that says which people your result is compared with.

Check 1: who signs it

The Standards open their chapter on documentation with a principle that applies to every seller: in the wording of Standard 7.0, information relating to tests should be clearly documented so that those who use tests can make informed decisions about which test to use, how to administer it and how to interpret scores. Documentation has an author. When you can see who wrote it, you can do what no badge allows: search for that person or team independently of the site, find their other work, and check any credential they claim against the body that issues it.

A named author proves accountability and nothing more. A license belongs to a person and says nothing about whether this particular test was built well, and a long list of titles can sit on a test with no manual. The reverse holds too. A pseudonym or a company name is not a failure by itself, since many test builders are teams and many publishers are institutions, and when a name gives you less to trace, put more weight on checks 3 and 4. Established publishers show what the full version looks like: the Pearson page for the WAIS-5 lists a qualification level for buyers and a technical and interpretive manual, and Mensa is an organization whose testing page says how each route is administered.

One caution for this page itself. Because the page you are reading is published by a seller, its own About page is part of the evidence, and the ACIS section further down shows what the site publishes.

Check 2: whom the score is compared with

An IQ score is a position in a reference group, so a score with no described group has no meaning. Standard 5.8 puts it directly: norms, if used, should refer to clearly described populations. Standard 7.2 asks the publisher to document the population the test is intended for and, when norms are provided, to explain how the data were gathered, describe the norming population by relevant demographic variables and report the years in which the data were collected. Those three requests give you a one minute test. Find the sentence that names the group, check that the age range covers you, and look for a year.

Wording matters. A percentile reported among "all test takers" on a site is a comparison with that site's own users, who are a self-chosen set of people and not the adult population. For a purely illustrative case, which is our arithmetic and not a finding from any study, suppose a site's users average 5 points above the adult population on the same scale; a person who sits exactly at the adult average would then be reported 5 points below that site's average, and the percentile would shift with it. The score is not wrong in either case. It answers a different question, which is why the page on how IQ scores are normed matters before you compare numbers from two sites.

A pass on check 2 is limited too. A stated age range of 16 to 90 tells you whom the score claims to cover and says nothing about how well each age band was sampled. That belongs to the manual, which is check 3. For a reader who wants an adult-normed picture specifically, the entry page on IQ tests for adults explains what to look for in the age range.

5 Checks 3 and 4: Is There a Technical Manual, and Has Anyone Outside the Seller Checked It?

A technical manual lets you inspect the evidence and independent studies let you doubt it, and the two answer different questions: the first asks whether the seller has shown its work, the second asks whether anyone else has looked. Treat a missing manual as a larger warning than a missing independent study, because documentation is something a seller can always publish.

Check 3: a technical manual

What a manual should contain is not a matter of taste. The Standards say in Standard 7.4 that test documentation should summarize test development procedures, including the statistical analyses used, evidence of the reliability or precision of scores and the validity of their recommended interpretations, and the comment adds that standard errors of measurement and normative data should be provided as appropriate. Standard 2.3 adds that reliability estimates should be reported for each total score, subscore or combination of scores that is to be interpreted, because a total can be precise while a subscore is not.

The American Psychological Association gives the same advice to its own members. Guideline 6 of the APA Guidelines for Psychological Assessment and Evaluation, approved by the APA Council of Representatives in March 2020, says that psychologists endeavor to select tools that demonstrate sufficient validity evidence for their uses, sufficient score reliability and sound psychometric properties. The steps it lists include reading the publisher's technical manual and reviewing the design and the standardization sample, and also reading reviews and research studies. In other words, the check you are about to run is one that the APA suggests psychologists consider.

Reading a manual in five minutes means looking for five things: a description of the norm sample, reliability for every score you will actually read, the standard error, validity evidence beyond the test agreeing with itself, and an account of how raw performance becomes a reported score. Two cautions apply. A manual is the publisher's account, so its existence shows that the evidence can be scrutinized and does not show that it survives scrutiny. And a manual written for specialists is not a failure; it is a different document from the plain language report, and you need both. The ACIS manual is published on the technical manual page, and this page quotes no figure from it on purpose: read it yourself, and apply the five items above.

Check 4: independent evidence

Standard 7.3 asks that test documents cite a representative set of studies about general and specific uses of the test, and that unpublished summaries be available on request. The words to look for are studies by people other than the seller, and reviews that are not written by a competitor or an affiliate. Peer review is a filter and not independence: the paper that introduced the ICAR, Condon and Revelle (2014) in Intelligence, was written by the researchers who developed it and published after review, and the ICAR page hosted by TU Dortmund University is open about what the resource lacks, namely norms. That combination, open documentation with an honest statement of limits, is what a good result on checks 3 and 4 looks like even when the test cannot give you an IQ.

Many perfectly usable tests have little independent evidence, so the absence of it should move you only a notch. The practical rule is to weigh it against price and purpose. Paying a small fee for a personal estimate while knowing that outside studies are thin is a different decision from paying for a score you plan to hand to someone else. For the terms a manual uses, the page on reliability and validity defines them with examples.

6 Checks 5 to 7: What Does It Cost, What Is Free, and What Does the Refund Say?

Checks 5 to 7 are about money and consent, and they are the easiest to verify because, in the United States, the law and the regulator say what a seller of recurring charges must show before taking your card. In the United States a main federal law for online sales that renew is the Restore Online Shoppers' Confidence Act. The Federal Trade Commission's staff report Bringing Dark Patterns to Light, dated September 2022, summarizes it: a seller using a negative option feature, meaning charges that continue unless you cancel, must clearly and conspicuously disclose all material terms before obtaining your billing information, obtain your express informed consent before charging you, and provide a simple mechanism to stop recurring charges. This page is general information and not legal advice.

Check 5: price and billing before the first item

The test is concrete. Before the first question, can you read the total price, whether it is one payment or a repeating charge, and what the trial turns into when it ends. The report describes why this matters with an experiment: a workshop panelist presented research in which one group was told about the automatic monthly charge and another was told only that terms and conditions apply, with the charge in small gray text, and more than twice as many consumers in the hidden information group accepted the free trial. A separate academic study, Mathur and colleagues (2019), analyzed about 53,000 product pages from about 11,000 shopping websites and discovered 1,818 dark pattern instances of 15 types in 7 broader categories. Both concern online selling in general and say nothing about any IQ site in particular, but they show the pattern is measurable, which is why the check is worth a minute.

For a calibration point, Mensa's testing page lists prices of 60 and 99 dollars and says promotional discounts appear in the cart at checkout. ACIS shows one time prices on its home page with no subscription, and those prices can change, which is why every price on this page carries a reading date. If you want the billing models explained, the paid IQ test page linked earlier sets them out. ACIS also publishes dated reviews of other products. The MyIQ review, the Brainable review and the IQTest.com review each set out what that product's public pages showed in August 2026 about price, billing and claims, as far as the reviewers could read them. Each states that it comes from the team behind ACIS, which sells a competing assessment, so weigh it accordingly and open the pages it cites.

Check 6: what is free and what is paid

A clear line between the free part and the paid part is a courtesy and a diagnostic. What you want to know before starting is what the free part returns, whether a result is shown at all and what the payment unlocks. Asking for an email address is not a warning sign by itself; not knowing why it is asked is.

There is one more test you can run yourself, and it is our own suggestion and not a finding from research. A score should move with your answers. On a free part, and only where the site's terms allow it, a deliberate low effort on a few items and a serious attempt on the same kind of items should not return the same number. If a tool reports an equally high result whatever you do, the number carries no information about you. Do not pay to run this test on a paid product. The reasoning behind it is simple: a score is a measurement only if it can vary with the thing measured. The page on taking an IQ test without paying covers the free options in detail, and ACIS offers a free trial with no card for readers who want to see the format first.

Check 7: the refund terms

Read the refund terms before paying and look at three things: who decides, the window, and the conditions. A phrase such as "money back guarantee" without those three is a headline and not a term. The ACIS guarantee reads, literally: "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." That is a conditional guarantee, not an unconditional one, and it does not promise a refund if you dislike your score.

For recurring charges, the FTC report says the simple mechanism should be at least as easy to use as the method you used to sign up, in the same medium, and that sellers should not subject you to new offers or similar attempts to save the account that impose unreasonable delays on your cancellation. If you are already in a charge you did not expect, the sibling page on how to cancel an IQ test subscription takes the steps in order.

7 Checks 8 to 10: Does the Score Show Its Uncertainty, Its Limits and Its Conditions?

The last three checks read the report and the fine print for honesty about what a number is: an estimate with an error band, for a stated use, produced under stated conditions. A seller that is candid on these three points is telling you how to avoid over-reading the result.

Check 8: the score arrives with its uncertainty

Standard 6.10 says that when scores are released, the interpretation should describe in simple language what the test covers, what scores represent, the precision or reliability of the scores and how they are intended to be used, and its comment notes that score precision might be depicted by error bands showing the standard error of measurement. Standard 2.14 adds that conditional standard errors should be reported at several score levels when possible. On a report, this looks like a range around the headline number. A single figure with no band leaves you unable to judge how precise it is.

The practical consequence is large. An IQ report with a 95 percent confidence interval is telling you the range in which your underlying score plausibly lies, so a difference of a point or two between you and a friend, or between two sites, is not a meaningful difference. The page that shows what such a report contains is the IQ test report example, and the page on standard deviation 15 explains the scale the band is drawn on.

Check 9: the stated limits of use

Standard 7.1 asks publishers to document the recommended uses of a test and to caution against misuses that can reasonably be anticipated. A seller that claims its number can be used for anything is either not paying attention or is selling something other than a test. Look for a sentence that says what the score is not for. The common overreach has a name, the certificate: a document that looks official and that no receiving party has agreed to accept. The page on the IQ test certificate takes up that question. The rule on this page is simpler. A certificate shows that a seller issued a document. Whether anyone accepts it is a question for the receiving party, and the Mensa testing page shows how specific that can be: it states the exam is for admission to Mensa and not to quantify intelligence.

Check 10: the conditions of administration

The International Test Commission distinguishes four modes of delivery in its International Guidelines on Computer-Based and Internet-Delivered Testing (International Test Commission, 2006). In open mode there is no direct human supervision and no means of authenticating the test taker. In controlled mode there is no supervision but the test is available only to known test takers. In supervised mode a person supervises and the identity of the test taker can be authenticated. In managed mode there is a high level of human supervision and control, normally in dedicated testing centers. An online IQ battery taken at home is therefore in open or controlled mode, and for computer-generated score reports the same guidelines tell users to be aware of the case where interpretations are based on proctored, standardized conditions, the test was given in open or controlled mode and no evidence supports the report under those conditions.

The Standards say the same thing in two places. A comment on Standard 3.4 warns that unproctored administrations where standardization cannot be ensured could give some test takers an advantage over others, and that in such situations test results should be interpreted with caution. And Standard 1.10 asks that the conditions under which validity data were collected be described in enough detail that users can judge their relevance; its comment gives unproctored online testing against proctored on site testing as an example. What you want from a seller is a plain statement of the conditions and of what the test expects of you. The guidelines say on screen information should cover at a minimum the number of items, the timing and the types of items, and the testing procedure.

8 What Should a Legit IQ Test Measure?

A legit IQ test says which abilities it samples, in names that researchers recognize, and it samples more than one, because a general score built from a single kind of task is a narrower thing than its label suggests. The recognized vocabulary is the Cattell-Horn-Carroll (CHC) framework, the label for the combination of the Cattell-Horn Gf-Gc model and Carroll's three-stratum model, discussed in McGrew's 2009 article in Intelligence on CHC theory and the Human Cognitive Abilities project. In that framework the broad abilities have short names: Gf for fluid reasoning, Gc for comprehension and knowledge, Gq for quantitative knowledge, Gv for visual processing, Gwm for working memory and Gs for processing speed.

That vocabulary is a check in itself. A site that reports a "logic IQ" or a "memory IQ" without saying which established ability the tasks sample leaves you unable to compare the result with any research. The pages on the CHC model and on what IQ measures lay out the framework and its limits.

Take ACIS as a worked example of how a battery states its coverage. It has 20 subtests in six CHC domains, reported as a Full Scale IQ and six indices: verbal comprehension (VCI), fluid reasoning (FRI), quantitative reasoning (QRI), visual spatial (VSI), working memory (WMI) and processing speed (PSI). A fluid reasoning subtest such as Matrix Reasoning and a working memory subtest such as Digit Span feed different indices, and the report shows each separately. The Quick form uses 6 subtests in 3 domains, the Optimized form 13 in 5 domains and the Full Scale form all 20 in six domains. Fewer subtests mean fewer abilities covered and, as the next section explains, usually a wider band of uncertainty.

9 How Do You Read a Score From a Test That Passes the Checks?

A score from a test that passes is still read as a range inside a described group, with precision that depends on how many tasks sit behind it. The conventions are shared across the major batteries. ACIS reports scaled subtest scores from 1 to 19, where 10 is the average and the standard deviation is 3, and reports indices and the Full Scale IQ on the familiar scale where 100 is the average and the standard deviation is 15, along with percentiles and a 95 percent confidence interval. The percentile is the share of the reference group at or below your score, and it changes with the group, which is the point of check 2. Two scores on different scales cannot be averaged by eye, a trap the page on IQ scores and percentiles sets out.

Precision tends to fall as the test gets shorter. The Standards state in their chapter on reliability that if an assessment is shortened by decreasing the number of items or tasks, reliability is likely to decrease, and that lengthening it with comparable tasks is likely to increase it. Standard 2.3 makes the same point at the level of scores: a total can be precise while a subscore is not, so each score you read needs its own reliability evidence. For you, that translates into a plain rule. One subtest is less precise than an index, an index is less precise than a battery, and the interval on the report is the quickest way to see by how much.

That is also why this page gives no ranges for ACIS or for any other test. The width of the interval is a feature of each test and each form and belongs on that test's report and manual, where you can check it.

10 What Does an Unsupervised Session Need?

An unsupervised session works only if you supply the conditions a proctor would, because the score assumes them and nobody else will check. The International Test Commission guidelines say test takers should be informed in advance of the procedures and asked to confirm that they will complete the tests according to instructions, for example by not seeking assistance and not colluding with others. Treat that as a promise to the measurement, not to the site.

The practical list is short. Use a device and browser you are comfortable with, in a quiet room, and finish each subtest in one sitting. Three ACIS subtests, Digit Span, Alphanumeric Sequencing and Arithmetic, present their items as audio, so working sound is required. ACIS allows breaks and saves progress between subtests, with 30 days to complete a form, so there is no reason to take a subtest when you are tired or interrupted. Do not search for items or ask anyone for help; either choice answers a different question from the one the score claims to answer. Finally, the ACIS manual says verbal and knowledge tasks are affected by English proficiency and educational exposure, and ACIS is in English only.

These conditions are the practical form of check 10. They are why an unsupervised score is read as an estimate, a position the Standards support when they say results from unproctored administration should be interpreted with caution.

11 What Does ACIS Publish on the Ten Checks?

This table shows what ACIS publishes on each of the ten checks, as read on November 4, 2026, so you can run them yourself.

Read the table against your purpose. If you want a number to inform your own decisions and you are willing to read the manual, these documents are the place to start. If your purpose needs more than a personal estimate, a supervised assessment is the right tool.

12 Where ACIS sits, and what a buyer should do

ACIS is a normed, adult, online battery with documentation, visible prices and a conditional guarantee, and because it is unsupervised and English only, the score it gives is a personal estimate and not a credential. The research page states the limit literally: ACIS "is not intended for clinical diagnosis, hiring decisions, accommodation determinations or admission to high IQ societies." That sentence is the boundary. Nothing on this page says ACIS is accepted by a society or an employer, that it predicts a result on any other test or that it is endorsed by any organization.

The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014) supply the frame for using any result. They ask publishers to document the intended population, the norming procedures and the evidence (Standards 7.2 and 7.4), they say unproctored administrations call for cautious interpretation (a comment on Standard 3.4), and they put the responsibility on the user who applies a score to a purpose with little validity evidence to document why and to collect the evidence (Standard 9.4). The APA Guidelines for Psychological Assessment and Evaluation (2020) ask psychologists to choose tools with sufficient validity evidence, sufficient score reliability and sound psychometric properties (Guideline 6). Read together, they say that a score is only as good as the match between its documentation and your purpose.

A concrete way to use that:

1. Run the ten checks on every site you are considering, ACIS included, in the table above. Note where each fails and weigh them as the sections explain.
2. Open the technical manual of any test you plan to pay for, and read it for the five items in check 3. For ACIS that is the document published at the technical manual page.
3. Match the score to the purpose. For personal curiosity, a normed online score read with its confidence interval is reasonable. For a clinic, a hiring process, a school or a society such as Mensa, it is the wrong tool, and the official routes named earlier are where to look.
4. Start free before you pay. The ACIS trial needs no card, and you can see the format and the example report before choosing a form. Prices were read on November 4, 2026 and can change.
5. Check the community with care. Forum recommendations are opinion and not independent evidence; the sibling page on the best IQ test according to Reddit shows how to judge one against published criteria.

13 Sources Behind This Page

Every web page below was opened on November 4, 2026, and every DOI was checked against Crossref on the same day. The ACIS statements in the text come from the published ACIS site as read on November 4, 2026, and are not listed as sources about intelligence. No competing product is linked or used as a source.

• American Educational Research Association, American Psychological Association and National Council on Measurement in Education (2014). Standards for Educational and Psychological Testing. Washington, DC: American Educational Research Association. Open access edition.
• International Test Commission (2006). International Guidelines on Computer-Based and Internet-Delivered Testing. International Journal of Testing, 6(2), 143-171. doi.org/10.1207/s15327574ijt0602_4.
• American Psychological Association, Task Force on Psychological Assessment and Evaluation Guidelines (2020). APA Guidelines for Psychological Assessment and Evaluation. Approved by the APA Council of Representatives, March 2020. apa.org.
• Federal Trade Commission, Bureau of Consumer Protection staff (2022). Bringing Dark Patterns to Light. Staff report, September 2022. ftc.gov.
• Mathur, A., Acar, G., Friedman, M., Lucherini, E., Mayer, J., Chetty, M., and Narayanan, A. (2019). Dark Patterns at Scale. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), 1-32. doi.org/10.1145/3359183.
• Condon, D., and Revelle, W. (2014). The international cognitive ability resource: Development and initial validation of a public-domain measure. Intelligence, 43, 52-64. doi.org/10.1016/j.intell.2014.01.004.
• McGrew, K. (2009). CHC theory and the human cognitive abilities project: Standing on the shoulders of the giants of psychometric intelligence research. Intelligence, 37(1), 1-10. doi.org/10.1016/j.intell.2008.08.004.
• International Cognitive Ability Resource project page, TU Dortmund University (read November 4, 2026). sms.statistik.tu-dortmund.de.
• Pearson Assessments (read November 4, 2026). Wechsler Adult Intelligence Scale, Fifth Edition, product page. pearsonassessments.com.
• American Mensa (read November 4, 2026). Mensa testing page. us.mensa.org/join/testing.
• University of South Florida Psychological Services Center (read November 4, 2026). Fees, Clinic Policies. usf.edu.

14 Frequently Asked Questions