How accurate are online IQ tests?

Online IQ tests vary widely in quality. The best ones — typically those based on the ICAR public-domain catalog or the items released by the Open-Source Psychometrics Project — provide useful estimates of cognitive ability with documented psychometric properties. The standard error of measurement on a brief online screener is roughly ±7 IQ points, compared to ±3 for a clinically administered WAIS-IV. So a measured score of 115 on an online test means 'somewhere between about 101 and 129 with 95% confidence' — useful for self-curiosity but not precise enough for any high-stakes application. Tests that report exotic-sounding scores like 'IQ 168' from a 10-question quiz are not measuring what they claim to measure; ignore them. Reputable online tests will be transparent about their item sources, scoring approach, and measurement error.

This question comes up frequently from users of free online IQ tests and from people considering whether to pursue a clinical evaluation. The full answer depends on context — what the score will be used for, how recently the test was administered, and what other information is available. The brief answer above captures the broad consensus from the published research literature; the linked deep-dive articles cover the underlying evidence in more detail.

Related considerations include the standard error of measurement on the relevant test, the population the test was normed against, and the specific cognitive abilities the test samples. A score is much more informative when interpreted alongside these contextual variables than when reported as a bare number.

If this answer raises further questions, see the related FAQ entries listed in the sidebar and the longer-form articles on the same topic in the article library. The site is designed to provide layered depth: the FAQ entries offer concise answers, the deep-dive articles offer the underlying research, and the score-interpretation pages tie the abstract concepts to specific result bands.