IQ score ranges and percentiles

Last reviewed 6 September 2026

The scale is built, not discovered

An IQ score is not a count of anything. There is no unit of intelligence being tallied up. What a modern IQ score reports is a rank: where your performance sits within a reference distribution of other people's performance, expressed on a scale that has been deliberately fixed so the average is 100 and the standard deviation is 15.¹

That choice is a convention, not a discovery. Test publishers standardise a test on a large sample chosen to represent a population, look at the raw score distribution, and then stretch and shift it onto the 100/15 scale. This is what "deviation IQ" means: your score expresses how many standard deviations you sit from the mean of your age-matched norm group.

Two consequences follow immediately, and they explain most of the confusion around these numbers:

• The average is 100 by construction. It cannot be otherwise. If the population got uniformly better at the test tomorrow, the average would still be reported as 100 after renorming.
• Scores are only meaningful relative to a norm group. "IQ 120" is shorthand for "performed better than about 91% of the people in the reference sample". Change the reference sample and the number changes.

Full score, percentile and rarity table

Percentiles below are computed from the normal distribution with a mean of 100 and a standard deviation of 15. "Percentile" is the share of people expected to score at or below that value; "rarity" is roughly how many people you would need before expecting to find one scoring that high or higher.

The middle of that table is where almost everyone is. About 68% of people fall between 85 and 115, and about 95% between 70 and 130. The interesting-sounding scores are, by construction, the ones almost nobody has.

Classification labels, and their limits

Test manuals attach descriptive labels to score bands. These are useful shorthand and are widely misused, so it is worth being precise about what they are.

Why rarity accelerates at the extremes

Look again at the rarity column. Going from 100 to 115 takes you from 1 in 2 to about 1 in 6. Going from 130 to 145 takes you from roughly 1 in 44 to roughly 1 in 741. The same 15-point step costs vastly more the further out you go.

That is just the shape of the normal curve: the tails thin out exponentially. It has one very practical implication. Near the extremes, small differences in score correspond to large differences in rarity - which is exactly where measurement error is largest and where there are fewest items to distinguish people. Precise-sounding claims about very high scores are the least trustworthy numbers in the whole field, and any test that hands out 150s freely is telling you something about the test rather than about you.

The part most tables leave out

Every table like the one above presents scores as exact. They are not. A score is an estimate with a standard error of measurement attached, and a responsible report states it.

Even the WAIS-IV, administered one-to-one by a trained examiner over an hour or more, has a standard error of about 2.2 points on Full Scale IQ - so a reported 100 carries a 95% confidence interval of roughly 96 to 104.¹ A good short online test is considerably less precise: the test on this site reports an interval of about ±9 points near the middle of the range, and says so on the result page.

The practical rule: read any single IQ score as a band, not a point. If two scores differ by less than about 10 points on a short test, you have not learned that one person is higher than the other.

A note on SD 15 versus SD 16

Most modern tests use a standard deviation of 15, but not all. Older Stanford-Binet forms and some high-range tests use 16, and Cattell's scales used 24. The same underlying performance produces different numbers on each scale, which is why comparing a score from one test against another without noting the metric is meaningless.

At +2 SD, an SD-15 test reports 130 and an SD-16 test reports 132. At +4 SD the gap widens to 160 versus 164. If someone quotes a very high IQ without naming the test and the scale, the number does not mean anything specific.

Common questions

References

1. Wechsler, D. (2008). Wechsler Adult Intelligence Scale - Fourth Edition: Technical and Interpretive Manual. Pearson.
2. American Educational Research Association, American Psychological Association, & National Council on Measurement in Education (2014). Standards for Educational and Psychological Testing. AERA.
3. Roid, G. H. (2003). Stanford-Binet Intelligence Scales, Fifth Edition: Technical Manual. Riverside Publishing.
4. Gottfredson, L. S. (1997). Why g matters: The complexity of everyday life. Intelligence, 24(1), 79-132.
5. Deary, I. J., Strand, S., Smith, P., & Fernandes, C. (2007). Intelligence and educational achievement. Intelligence, 35(1), 13-21.

Related reading