Most free IQ sites tell you a number and nothing else. This page explains exactly how CheckIQFree produces yours: where the questions come from, how raw answers become a score, what that score is compared against, and where the method is currently weak. If any of it looks wrong to you, tell us and we will fix it.
There is no unit of intelligence the way there is a unit of length. An IQ score only means something relative to a reference group. By long-standing convention that scale is centred on 100 with a standard deviation of 15, which puts roughly two thirds of people between 85 and 115, and about 2 in 100 above 130.
So the honest way to read your result is not "I have 116 units of intelligence" but "on this particular set of reasoning puzzles, under this time limit, I placed above about 86 out of 100 people."
Every question in our IQ tests is generated by software we wrote and own. Nothing is copied from another test publisher.
The figural items are 3×3 matrices. A generator first picks the rules, then draws the grid to fit them. The rules are the standard ones described in the research literature on matrix reasoning:
Difficulty is simply how many of these rules operate at once, from one in the easiest items to four in the hardest.
The number and letter items are generated the same way: a rule is chosen first (arithmetic step, geometric ratio, constant second difference, two interleaved series), then the sequence and the distractors are built from it.
Because the answer is determined before the picture exists, every item has exactly one defensible solution. We also check every generated item automatically at the pixel level to confirm no two answer options render identically — an item where two choices look the same is unfair regardless of the underlying logic.
We convert the proportion of items you answered correctly into a position on the standard scale:
IQ = 100 + 15 × (p − mean) ÷ SD
where p is your proportion correct, and mean and SD are the expected proportion correct and its spread for that specific test. Results are reported between 55 and 145; a short online test cannot meaningfully distinguish scores outside that range.
For the 26-question International Standard test, currently using mean = 0.50 and SD = 0.18, the conversion is:
| Correct | Score | Percentile |
|---|---|---|
| 4 of 26 | 71 | 3rd |
| 8 of 26 | 84 | 14th |
| 10 of 26 | 90 | 25th |
| 13 of 26 | 100 | 50th |
| 16 of 26 | 110 | 75th |
| 18 of 26 | 116 | 86th |
| 21 of 26 | 126 | 96th |
| 23 of 26 | 132 | 98th |
| 26 of 26 | 142 | 99th+ |
The High Range test uses harder items, so its expected proportion correct is lower (0.40) and the same raw count maps to a higher score. That is why scores are only comparable within the same test.
A properly normed test derives its mean and SD from a large sample of real test-takers. Ours are currently set from the difficulty design of the item generator, not measured from data, because the question bank is new.
Our commitment: once each test has collected at least 500 completed attempts on the current question bank, we will replace these figures with the observed mean and standard deviation from those attempts, publish both the old and new values on this page, and note the date of the change. Until that happens, treat scores here as a reasonable estimate rather than a measurement.
Between February 2025 and August 2026 this site used a different formula: seven points per correct answer, starting from zero. That produced results that do not exist on any real IQ scale — a perfect run on the 26-question test returned 182 — and it placed the median test-taker at 70.
Across 5,898 completed tests in that period the mean result was 69. Those numbers were wrong, and anyone who took a test here before 22 August 2026 received a score that should be disregarded. Those records have been marked in our database as belonging to the old scale and are excluded from the public leaderboard rather than deleted.
We are describing this openly because a scoring page that only lists a site's strengths is not worth reading.
You can take any test and see your result without entering a name, an email address, or anything else. Adding your score to the public leaderboard is optional and happens after the result is shown, never before. See our Privacy Policy for details.
CheckIQFree is published by GPWEBMEDIA Co., Ltd, a company registered in Viet Nam (business registration number 4001291359, issued 13 January 2025 by the Department of Planning and Investment of Quang Nam Province), with its registered office at 03 Ly Dao Thanh, Hoa Thuan Ward, Tam Ky City, Quang Nam, Viet Nam. Our company website is gpwebmedia.com.
We are a web development company, not a psychology practice. What we contribute is the software: the item generator, the scoring implementation, and the platform. Interpretation of psychometric content on this site is reviewed separately, and reviewer credentials are listed on the pages concerned.
This site carries advertising through Google AdSense. Advertisers have no influence over test content, scoring, or what we publish here.
Questions or corrections: contact us. We would rather be told we are wrong than leave an error in place.
Last reviewed: 22 August 2026.