Free Raven's Progressive Matrices Practice Test

What a Raven's-style test measures

This practice test targets fluid intelligence, what John C. Raven called eductive ability: the capacity to make sense of confusing information by working out the relationships inside it, with nothing memorised to fall back on. Each item shows a pattern with one piece missing, and you infer the rules that govern how the pattern changes across rows and columns, then apply them to pick the piece that completes it. The rules stack as the test goes on, one transformation at the easy end, two or three interacting at the hard end.

Raven built the original in 1936, and the published versions, Standard, Coloured, and Advanced Progressive Matrices, remain among the most heavily researched instruments in psychology precisely because they are so strongly g-loaded: performance on them tracks general intelligence more closely than almost any other single test. They are also non-verbal by design, which makes them useful across languages and education levels where a vocabulary-based test would not be comparable.

This version presents the same logical relationships in text rather than as visual grids, so it loads in any browser and works with a screen reader. The reasoning operation, find the rule, apply the rule, is identical; only the input format changes. It is inspired by the format and is not affiliated with or endorsed by Pearson, the publisher of the official assessments.

How to read your result

With only 12 items the bands are broad, so read your score as a range rather than a precise level. Developing (below about 42 percent, roughly 0 to 4 correct) usually means the multi-rule items are outrunning you; the single-rule items at the start are the model, and studying the explanations for the harder ones builds the skill fast. Average (about 42 to 66 percent, roughly 5 to 7 correct) is a solid mid-range result on abstract reasoning.

Strong (about 67 to 83 percent, roughly 8 to 9 correct) means you are tracking two or more interacting rules without losing one, which is the marker most associated with fluid intelligence. Excellent (about 84 percent and up, 10 to 12 correct) means you cleared the items where several transformations combine, a level that sits in the upper percentiles for this question type.

Because the test is short, a single wrong answer moves your band. If you are between bands, retake after a day rather than reading too much into one item.

How this differs from the other tests on IQPF

This is the most specific test on the site. Where the full 30-question IQ test blends abstract reasoning with verbal knowledge and memory, this isolates the language-free, knowledge-free core and nothing else. That makes it the best single test here for estimating fluid reasoning, and the least affected by vocabulary or schooling.

It is easy to confuse with the spatial reasoning test because both use figures, but the operations are opposite: spatial reasoning asks you to rotate or fold a specific shape, this asks you to infer an abstract rule that the shapes only illustrate. It is also the inductive counterpart to the logical reasoning test: there you apply rules that are given, here you discover rules that are not. People frequently score differently on those two, and that difference is genuine information about how you reason.