Short answer: The average IQ is 100. That is a convention, not a finding: publishers norm each test on a large sample and assign 100 to the middle performer, with 15 points per standard deviation. About two in three people score between 85 and 115. Because raw performance rose for most of the 20th century, a phenomenon called the Flynn effect, tests are renormed every decade or two to pull the average back to 100.
At a glance
| Average IQ | 100, fixed by norming |
|---|---|
| Standard deviation | 15 points on the Wechsler scales and the Stanford-Binet 5 |
| Two-thirds of people | 85 to 115 (about 68 percent) |
| Middle half of people | About 90 to 110 |
| 20th-century drift | About 3 points per decade (Trahan et al., 2014) |
| Newest adult norms | WAIS-5, data collected 2023 and 2024 |
| National averages | Disputed; built on uneven samples |
100 is a definition, not a discovery
An IQ score is a rank. A publisher gives a new test to a sample built to mirror the population, then converts every raw result into a standard score: 100 for the sample's middle, 15 points for each standard deviation above or below. The median performer receives 100 because the scoring rule says so.
The Wechsler Adult Intelligence Scale shows how the work is done. Its fourth edition, the WAIS-IV, was standardized in 2008 on 2,200 Americans aged 16 to 90, with the sample matched to US Census proportions. Its successor, the WAIS-5, was released in late 2024 on norms gathered in 2023 and 2024. Each sample defines what 100 means for that edition.
That is why the question "What is the average IQ?" has a fixed answer while the question "How smart is the average person?" does not. The number stays put. The performance behind it moves. For the mechanics of scoring, see how IQ is measured, and for the exact midpoint, IQ 100.
What counts as average
On a bell curve with a mean of 100 and a standard deviation of 15, the middle of the distribution is wide. More than a quarter of people land within five points of 100, and roughly half fall between 90 and 110, the band the Wechsler tests label Average.
| Score range | Share of people | Meaning |
|---|---|---|
| 95 to 105 | About 26% | Within a third of a standard deviation |
| 90 to 110 | About 50% | The Wechsler Average band |
| 85 to 115 | About 68% | Within one standard deviation |
| 70 to 130 | About 95% | Within two standard deviations |
A 95 and a 105 are statistically close. Both sit inside the error band of a single test session, which is roughly five points either way. See IQ 95 and IQ 105 for the neighbors, or the IQ rarity chart for every band.
The Flynn effect: the average kept climbing
Raw scores rose across most of the 20th century, and fast. The researcher James Flynn documented the gains in two papers in Psychological Bulletin: "The Mean IQ of Americans: Massive Gains 1932 to 1978" in 1984 and "Massive IQ Gains in 14 Nations" in 1987.
The scale of the drift is now well measured. A 2014 meta-analysis by Lisa Trahan and colleagues estimated about 2.93 points per decade on Stanford-Binet and Wechsler tests. Dutch military conscripts gained 21 points between 1952 and 1982, about seven a decade. A larger meta-analysis by Jakob Pietschnig and Martin Voracek in 2015, covering nearly four million people tested between 1909 and 2013, found annual gains of 0.41 points on fluid measures and 0.21 on crystallized ones, with the pace slowing in recent decades.
The causes remain debated. Candidates with published support include better nutrition, more schooling, iodized salt, lower lead exposure and a modern environment that rewards abstract thinking. Stuart Ritchie and Elliot Tucker-Drob estimated in 2018 that each extra year of formal education adds one to five points.
Renorming: resetting the scale
Rising raw scores create a problem for any test with fixed norms. A person scored against a 20-year-old sample is compared with a weaker field, and the result runs high. Publishers answer by renorming. The Wechsler Intelligence Scale for Children, first issued in 1949, was revised in 1974, 1991, 2003 and 2014; each revision reset its average to 100.
The gap between old and new norms is itself the evidence for the Flynn effect: today's test takers, scored on yesterday's tables, average well above 100. It also explains a practical rule for psychologists. A score is only as current as the norms behind it, so the edition matters. The WAIS-5 data, reported by Winter and colleagues in 2024, suggest the drift has slowed to about 1.2 points per decade rather than the three that earlier editions showed.
For the tests themselves, see the WAIS-IV explained and the WISC-V explained.
Where the average has started to slip
In parts of Northern Europe, the climb has stopped or reversed. Thomas Teasdale and David Owen tracked Danish conscripts and found gains of about three points per decade from 1959 to 1979, smaller gains through 1998, then a decline from 1998 to 2004. Norwegian conscript scores stopped rising after the mid-1990s, according to Jon Martin Sundet and colleagues in 2004.
The most cited study is Bernt Bratsberg and Ole Rogeberg's 2018 paper in PNAS. Using Norwegian military records, they found the rise and the later fall both appeared within families, among brothers born years apart. Because brothers share parents, the authors read the pattern as environmental rather than genetic.
None of this changes the published average. Each new edition still sets its middle at 100. What changes is how much raw reasoning that 100 represents.
Why national averages are contested
Lists of "average IQ by country" trace mostly to datasets compiled by Richard Lynn and Tatu Vanhanen, published as books in 2002 and 2006. Those numbers are not norms in the sense above. They are assembled from many different tests, given in different decades, to samples of uneven size and representativeness, then placed on one common scale.
The criticism is specific. Heiner Rindermann noted in 2007 the mix of tests and the "not always clear representativeness of the samples." Earl Hunt and Robert Sternberg argued in 2006 that a national figure means little without probability samples. Jelte Wicherts, Conor Dolan and Han van der Maas reported in 2010 that Lynn and Vanhanen excluded studies without published criteria; restoring them moved the sub-Saharan African estimate from 67 to 82. In 2020 the European Human Behaviour and Evolution Association recommended against using the dataset, and in 2024 seven researchers writing in the health news outlet STAT called on journals to retract Richard Lynn's articles.
National figures should be read as rough, contested estimates heavily shaped by schooling, health and the Flynn effect. Our average IQ by country guide sets out the numbers and their limits.
Where an individual score fits
For one person, "average" is a range, not a point. A licensed psychologist reports a score with a confidence interval, usually at 95 percent, because no test measures without error. A 102 on a well-normed test is best read as somewhere in the high 90s to low 100s.
Age does not move the target. Adult and child tests compare each person with peers of the same age, so the average is 100 at 8 and at 80; average IQ by age explains why. If you want a quick sense of your own range, our free IQ test is a practice screener, not a clinical instrument; how accurate online IQ tests are explains the difference. For what counts as above average, see what is a good IQ score, and the full library sits in our IQ guides.
Frequently asked questions
What is the average IQ score?
The average IQ is 100 on every modern, professionally normed test, including the Wechsler scales and the Stanford-Binet 5. Test makers set the middle of their standardization sample at 100 and use 15 points per standard deviation, so about 68 percent of people score between 85 and 115.
Is the average IQ 100 in every country?
On a test normed in a given country, yes: the national norming sample is set to average 100. Cross-country lists that show other averages pull results from many tests onto one common scale. Those compilations, mainly by Richard Lynn and Tatu Vanhanen, are disputed for uneven sampling and mixed tests, and critics including Wicherts and colleagues (2010) found their estimates shifted substantially when excluded studies were restored.
Is an IQ of 110 above average?
Slightly. A 110 sits at about the 75th percentile, at the top edge of the Wechsler Average band of 90 to 109 and the start of High Average. Given a typical error of about five points, a single 110 overlaps with scores that are plainly average. See IQ 110 for the details.
Has the average IQ gone down?
The published average cannot fall, because each new norm sets it at 100. Raw performance is another matter. It rose about three points per decade through much of the 20th century, then stalled or fell in Denmark, Norway and some other countries from the 1990s. Data from the WAIS-5, released in 2024, point to slower gains of about 1.2 points per decade in the United States.
Why do IQ tests need to be renormed?
Because raw scores drift. When a population's performance rises, people scored against an old sample look better than they are compared with their own generation. Publishers collect fresh norms every decade or two, as with the WISC revisions of 1974, 1991, 2003 and 2014, so that 100 again means the current middle.
Sources
- Flynn effect (Wikipedia) — Flynn 1984 and 1987; Trahan et al. 2014; Pietschnig and Voracek 2015; Teasdale and Owen; Bratsberg and Rogeberg 2018; WISC revision years; Winter et al. 2024
- Intelligence quotient (Wikipedia) — Median performance set to 100; mean 100 and SD 15; renorming and the negative Flynn effect
- Wechsler Adult Intelligence Scale (Wikipedia) — WAIS-IV 2008 sample of 2,200 aged 16 to 90; WAIS-5 2024 with 2023-24 norms
- Nations and intelligence (Wikipedia) — Criticism by Rindermann 2007, Hunt and Sternberg 2006, Wicherts et al. 2010; EHBEA 2020; 2024 call for retractions
- IQ classification (Wikipedia) — Wechsler Average band 90 to 109; 95 percent confidence intervals
- Psychological testing: test norms (Britannica) — Deviation IQ set for each age level
Editorial standard: IQ figures for historical figures are retrospective estimates and are labeled as such; scores for living people appear only when the person disclosed them or a reputable outlet reported them. Corrections: contact the editors.
