A single number summarising your cognitive ability is a strange and load-bearing thing. The IQ classification system is the layer of labels that makes that number legible — "Average", "Superior", "Very Superior" — and it carries a lot of weight in school placement decisions, occupational selection, and clinical assessment. This guide walks through what each band actually means, the percentile it represents, and what (and what not) you can read off it.
If you'd rather see your own score before reading the labels, our free 41-question IQ test maps you onto the standard Wechsler scale. If you want a direct answer to whether your score is "good" in practical terms, see what is a good IQ score?.
The shape of the IQ scale
The modern IQ scale is anchored by two numbers: a mean of 100 and a standard deviation of 15. Both are conventions, not discoveries. The mean is fixed at 100 because it's a round, easy-to-read centre point. The SD is 15 because that's the value Wechsler chose in 1939 and the field consolidated around it; older Stanford-Binet scales used SD 16, and a few research scales (Cattell's CFIT) use SD 24. Always check which scale you're reading.
Because IQ scores are forced to follow a normal distribution by the way they're scaled, the classification bands map onto rigid percentile cutoffs:
- One SD above the mean (115) is the 84th percentile. About 16% of people score higher.
- Two SD above the mean (130) is the 97.7th percentile. About 1 in 44 people score higher.
- Three SD above the mean (145) is the 99.87th percentile. About 1 in 740 people score higher.
- Four SD above the mean (160) is the 99.997th percentile. About 1 in 30,000 people score higher.
Going the other direction, the same cutoffs apply symmetrically below 100. This is why IQ score differences look small at the top and bottom of the scale but actually represent enormous shifts in rarity — the gap between IQ 130 and IQ 145 covers as much population as the gap between IQ 70 and IQ 130.
Where the classifications come from
The classification labels in current use are largely inherited from the original Wechsler-Bellevue Intelligence Scale (1939) and refined through successive WAIS revisions. David Wechsler chose verbal labels deliberately — he wanted clinicians to read scores in terms of cognitive functioning rather than as bare numbers, and the labels he picked have been carried through, with edits, into the WAIS-IV that most adult IQ assessments use today.
The Stanford-Binet (Terman, 1916, with multiple revisions) used overlapping but distinct labels — "Genius", "Very Superior", "Superior", "Average", "Dull Normal", "Borderline", "Defective" — many of which were later softened. "Defective" became "Mentally Retarded" in mid-century, then "Intellectually Disabled" in the 1990s, and the modern DSM-5 description avoids categorical IQ-defined labels entirely in favour of functional impairment criteria.
The classification words are not psychometric facts — they are clinical-cultural conventions, and they change as society reads them differently. The numeric cutoffs (mean 100, SD 15) have been remarkably stable since 1939. The labels attached to each band have shifted twice in the past 80 years and will shift again.
The standard classification ranges
Here is the WAIS-IV classification scheme, the most widely used in modern clinical and educational settings. Compatible labels from earlier Wechsler editions and from Stanford-Binet 5 are noted in parentheses where they differ.
| Score | Classification | Percentile | Population share |
|---|---|---|---|
| 130 and above | Very Superior (Gifted) | 97.7+ | ~2.3% |
| 120–129 | Superior | 91–97.7 | ~6.7% |
| 110–119 | High Average (Bright Normal) | 75–91 | ~16% |
| 90–109 | Average | 25–75 | ~50% |
| 80–89 | Low Average (Dull Normal) | 9–25 | ~16% |
| 70–79 | Borderline | 2–9 | ~6.7% |
| 69 and below | Extremely Low (Intellectually Disabled) | 0–2 | ~2.3% |
The percentile column describes where a score lies in the standardised sample. The population-share column gives the proportion of people inside that band. Together they capture the rarity of any given score: an IQ of 125, for example, is at roughly the 95th percentile and is held by about 1 in 16 people.
What each band predicts
An IQ classification is a probabilistic predictor, not a destiny. It captures average outcomes for a group of people scoring in that band, and individual variation within any band is enormous.
Average (90–109)
The middle 50% of the population. Comfortable with everyday verbal, numerical, and procedural reasoning. Most semi-skilled and many skilled occupations sit comfortably in this band. Standard high-school curricula are calibrated to be accessible across this range.
High Average and Superior (110–129)
Strong predictor of college completion, professional and managerial occupations, and reliable performance on tasks involving complex information. Schmidt and Hunter's 2004 meta-analysis of 85 years of personnel research found general mental ability is the single most predictive measure for job performance across roles, with the strongest correlations in cognitively demanding jobs.
Very Superior (130+)
Roughly the population from which graduate-school cohorts and STEM PhD programmes are drawn. Strong correlation with productive scientific output: Gottfredson (1997) and others have shown the 130+ band is overrepresented in research-faculty positions, patent-holding inventors, and high-creativity domains. The relationship is statistical — many people in this band do not pursue academic or technical careers, and many people in lower bands produce world-class work in their chosen field.
Low Average (80–89)
Successfully handles most daily-life tasks. May find purely abstract academic work harder to access without scaffolding. Schmidt & Hunter (2004) and Strenze (2007) show this band predicts somewhat lower socio-economic attainment on average, but the variance is wide enough that band-level prediction is poor at the individual level.
Borderline (70–79) and below
Functional impairment becomes more likely — not certain — in academic and complex occupational contexts. Modern diagnostic practice (DSM-5) avoids defining intellectual disability by IQ alone; the diagnosis requires both reduced cognitive scores and functional limitations in adaptive behaviour confirmed across multiple settings.
See where you land on the standard scale
41 questions · five cognitive domains.
The "genius" range and why it's blurry
"Genius IQ" is one of the most-searched phrases in this entire topic, and there is no precise psychometric answer to it. The label entered popular usage from early Stanford-Binet editions where Terman labelled scores above 140 "Near Genius or Genius". That label has been retired from every major modern test.
Three reasons the upper extremes are imprecise:
- Standardisation thins out. Most IQ tests are normed on samples of a few thousand people, which means there are very few data points above 145. Scores reported above that threshold are extrapolations, not direct measurements.
- Different tests give different numbers. Two well-validated tests can produce 10–15 point differences at the high end for the same person, because each test draws on a different mix of subscales.
- "Genius" isn't an IQ-test concept. Real-world creative output requires sustained motivation, domain knowledge, and a willingness to do unfashionable work for years. IQ correlates with these but does not predict them.
The honest summary: scores above 130 reliably indicate strong cognitive capacity. Scores above 145 indicate very rare cognitive capacity. Whether you call those "gifted" or "genius" or "highly gifted" is a labelling choice with no clean psychometric mapping.
The single biggest source of confusion when people compare IQ scores is that different scales use different standard deviations. Wechsler tests use SD 15, the current Stanford-Binet uses SD 15, but older Stanford-Binet editions used SD 16 and the Cattell Culture Fair III uses SD 24. The same person, at the same true ability, gets different numbers on each. Mensa accepts the top 2% of the population — that’s an IQ of 130 on Wechsler, but 132 on old Stanford-Binet (L-M), and roughly 148 on Cattell. None of those scores is "higher"; they're the same percentile on different rulers. Always check which scale a score is on before comparing it to anything.
Borderline and below
The lower bands deserve careful reading because they have historically been over-interpreted. Three things to keep in mind:
- Single test scores have wide error bands. A 95% confidence interval around a single WAIS-IV Full Scale IQ score is typically ±5 points. A score of 73 may indicate true ability anywhere from roughly 68 to 78.
- Functional adaptation matters more than the score. Modern diagnostic practice (DSM-5, AAIDD) requires evidence of impaired adaptive functioning across multiple domains in addition to a low score before any clinical label is applied.
- State factors influence the result. Anxiety, fatigue, language unfamiliarity, attention difficulties, and motivation all measurably depress IQ-test scores. A first low result should be retested under different conditions before any conclusions are drawn.
For an online 10-question or 41-question test, the same caveats apply at higher resolution: short tests are useful for self-assessment and ranging, not for clinical decision-making.
The limits of a single label
The classification system is convenient and the percentile cutoffs are rigorously defined. But the labels mask three realities that matter at the individual level.
Profiles vary across domains. A "Superior" overall score can mask weak working memory or weak processing speed under a strong verbal performance. The full IQ test scores you across multiple domains separately for exactly this reason — the headline number is less informative than the shape of the profile.
Test ceilings cap precision at the extremes. The WAIS-IV has a ceiling around 160 because most subtests run out of items at that level. Scores reported above 160 from any single test should be treated as approximations.
Labels are sticky. The classification you receive can shape how teachers, employers, and even you interpret your future cognitive performance. Carol Dweck's research on growth versus fixed mindsets shows that fixed labels can suppress effort and risk-taking, regardless of their accuracy. Use the band as a probabilistic data point, not as an identity.
How to read your own score honestly
- Note the test and edition. Different scales (Wechsler vs Cattell vs older Stanford-Binet) give different numeric mappings even for the same person. The same number means different percentiles on different scales.
- Read the band, not the digit. A score of 119 and a score of 121 are functionally indistinguishable given test error. Treat the band ("High Average" vs "Superior") as the meaningful unit.
- Look at the profile. Domain scores carry more practical signal than the single Full-Scale number. A flat profile reads differently from a spiky one even at the same total.
- Retest if the result feels off. Single-test performance is influenced by sleep, anxiety, illness, and motivation. A second sitting on a different day under good conditions is the cleanest read.
- Don't confuse rare with valuable. An IQ score predicts some outcomes (cognitively demanding job performance, academic completion) and not others (creativity, leadership, emotional intelligence, persistence). Use it for what it's calibrated to predict, and ignore the labels for everything else.
Frequently asked questions
What are the standard IQ classification ranges?
The Wechsler scale (WAIS-IV) classifies scores as follows: 130 and above is Very Superior, 120–129 is Superior, 110–119 is High Average, 90–109 is Average, 80–89 is Low Average, 70–79 is Borderline, and 69 and below is Extremely Low. The mean is set at 100 and the standard deviation at 15.
What IQ score is considered genius?
There is no clinical IQ classification labelled "genius". The term comes from older Stanford-Binet labels and tabloid tradition, not from modern psychometric practice. Scores above 130 (top 2%) are classified as Very Superior on the WAIS-IV, and scores above 145 are sometimes called "highly gifted", but the precise cutoff varies by test and edition.
What percentage of people have an IQ above 130?
Approximately 2.3% of the population scores above 130, by definition. The IQ scale is constructed so that 130 is two standard deviations above the mean, and on a normal distribution that places it at the 97.7th percentile. Scores above 145 (three standard deviations) are held by about 0.13% of the population.
What is the average IQ score?
The average IQ score is 100 by definition — the scale is constructed so that the mean of the standardisation sample equals 100 and the standard deviation equals 15. About 68% of the population scores between 85 and 115; about 95% scores between 70 and 130.
Are IQ classification ranges the same across different tests?
Mostly, but not exactly. The Wechsler scales (WAIS-IV) and Stanford-Binet (SB5) both use a mean of 100 and standard deviation of 15, so the bands align. Older versions of Stanford-Binet used a standard deviation of 16, and Cattell's CFIT uses 24. A score of 132 on WAIS-IV is not the same percentile as 132 on the Cattell scale — always check which scale is being used.
Related reading
- What Your IQ Score Actually Means — how the standard IQ scale works, and what your number does and doesn't predict.
- Average IQ by Age — how cognitive scores are normed across the lifespan and why the "average" stays at 100.
- IQ Test Sample Questions — six worked examples, one per cognitive domain, with answers and explanations.
- Pattern Recognition Test — the cognitive domain most tightly correlated with general intelligence.
- Working Memory Test — the single domain most strongly linked to fluid intelligence.
References
- Wechsler, D. (1939). The measurement of adult intelligence. Williams & Wilkins.
- Wechsler, D. (2008). Wechsler Adult Intelligence Scale — Fourth Edition (WAIS-IV) administration and scoring manual. Pearson.
- Roid, G. H. (2003). Stanford-Binet Intelligence Scales, Fifth Edition (SB5). Riverside.
- Terman, L. M. (1916). The measurement of intelligence. Houghton Mifflin.
- Gottfredson, L. S. (1997). Why g matters: The complexity of everyday life. Intelligence, 24(1), 79–132.
- Schmidt, F. L., & Hunter, J. E. (2004). General mental ability in the world of work: Occupational attainment and job performance. Journal of Personality and Social Psychology, 86(1), 162–173.
- Strenze, T. (2007). Intelligence and socioeconomic success: A meta-analytic review of longitudinal research. Intelligence, 35(5), 401–426.
- American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). APA.
- Dweck, C. S. (2006). Mindset: The new psychology of success. Random House.