It is the one question about IQ with a perfectly clean answer: the average is 100 because tests are built, and periodically rebuilt, to make it 100. The interesting facts hide underneath that definition — what raw ability does across a lifetime, what the Flynn effect did to a century of scores, and why the country-ranking tables floating around the internet deserve so little trust.
When a test is normed, thousands of people take it, and the scale is drawn so their average lands at 100 and their spread at 15 points per standard deviation. Years later, performance has drifted — so the publisher re-norms, and 100 quietly comes to mean a different raw performance. This is why comparing scores across decades, or across tests with different norming dates, misleads: the yardstick itself was redrawn.
Scores are age-normed, so the average stays 100 at every age — but the raw ability behind a 100 follows two very different curves. Fluid abilities — novel problem-solving, processing speed, spatial reasoning, working memory — peak between the early twenties and about thirty, then decline slowly and steadily; by the seventies, raw fluid performance runs one to two standard deviations below a 25-year-old's. Crystallized abilities — vocabulary, knowledge, comprehension — keep climbing to a peak around 45–54 and barely decline before roughly eighty.
Two practical consequences. First, an IQ test that ignores your age is scoring a 60-year-old's reasoning against 25-year-old benchmarks — one reason our test asks (optionally) for an age band and is untimed, since speeded tests amplify the age penalty. Second, "sharpness" in later life is real but selective: what fades is speed on novel puzzles, not accumulated judgment.
Half of everyone scores within 10 points of it — the distribution is that tight in the middle. The difference between IQ 98 and IQ 104 predicts almost nothing about two individuals; interests, conscientiousness and circumstance dwarf it. The scale earns predictive power toward its tails, and even there it stays one input among many. If you want to know where you sit, measure it — the reading is free and takes twenty minutes — and read the result as the range it honestly is.
Exactly 100, by construction. IQ tests are periodically re-normed so that the average person of each age group scores 100 with a standard deviation of 15. The average can never drift, because the scale is re-anchored to whatever the current average performance is — if everyone got smarter overnight, 100 would simply require more.
On the scale, 100 at every age — scores are computed within age groups. In raw ability the picture is a curve: processing speed and fluid reasoning peak around 20–29, working memory in the late twenties, while vocabulary and general knowledge peak around 45–54 and hold up well into the seventies. A 70-year-old and a 25-year-old who both score 100 are being compared to their own age peers, not to each other.
Raw scores rose for most of the 20th century — about 3 points per decade, the Flynn effect — which repeated re-norming kept hidden inside the constant 100. Since the 1990s several developed countries show the effect flattening or reversing slightly. Either way, the published average stays 100 because that is what re-norming does.
Treat country-ranking tables with deep suspicion. The best-known datasets mix tests from different decades, different instruments, tiny convenience samples, and estimates extrapolated from neighbouring countries. Serious psychometricians have documented severe quality problems with them. Cross-country comparisons of schooling-linked assessments (like PISA) rest on far better sampling — and even those measure schooling as much as anything.
It is the definition of typical — dead centre of the distribution, matching or beating half the population. Whether any score is "good" depends entirely on what it is for; see our full answer at what-is-a-good-iq.
Curious where you sit against the average? Untimed, free, instant honest results.
Take the free IQ testSee the full scale