IQ scores are built to follow a normal distribution: a mean of 100, a standard deviation of 15. That single design choice is why the bands below look the way they do — most people cluster near the middle, and the bands get rarer the further they sit from 100 in either direction. This guide walks through every band, the percentile each one represents, how the scale itself is constructed, and the interpretation mistakes that trip up almost everyone reading a score for the first time.
Percentiles above are approximate, derived from the standard normal distribution underlying IQ scoring, not measured directly from AurorIQ's user base.
How the scale is built
The "quotient" in intelligence quotient is a fossil. Early twentieth-century tests really did compute a quotient — mental age divided by chronological age, multiplied by 100 — which worked tolerably for children and collapsed entirely for adults, whose "mental age" stops climbing in any meaningful sense. Modern tests abandoned the ratio decades ago in favor of the deviation IQ: your raw performance is compared against a large, carefully constructed norming sample of people, and your score expresses how far you sit from that sample's average, in units of standard deviation.
The convention, set by David Wechsler's tests and adopted almost universally since, fixes the mean at 100 and one standard deviation at 15 points. So a score of 115 doesn't mean "15% better than average" — it means one standard deviation above the mean, which places you at roughly the 84th percentile. A score of 130 is two standard deviations up, around the 98th percentile. The points themselves have no absolute meaning; they're rulers laid over a population.
This is also why test publishers periodically re-norm their instruments. Raw performance on IQ-type tasks drifted upward across the twentieth century — the Flynn effect — so a fixed answer key would quietly inflate everyone's score over time. Re-norming re-centers the average at 100 for the current population, which means an IQ of 100 in 2026 and an IQ of 100 in 1956 describe the same relative position but not the same raw performance.
Every band, walked through
"Average" isn't a value judgment — it's a statistical description of where most of the population sits. Moving one band in either direction represents a real but modest statistical shift, not a different category of person. With that framing in place, here is what each band actually describes.
Below 70 — Low (bottom 2%)
Roughly 2.3% of the population scores below 70 on a well-normed test. This is the one band where a score can carry clinical weight — 70 is the conventional screening threshold in intellectual-disability assessment — but it's critical to understand that no diagnosis is ever made from a score alone. Clinicians require converging evidence about adaptive functioning in daily life, developmental history, and multiple assessments. A single low score, especially from an online test taken under unknown conditions, means very little by itself: illness, distraction, language barriers, and unfamiliarity with test formats can all suppress a score well below a person's actual ability.
70–79 — Below Average (2nd–9th percentile)
About 7% of people land here. Older literature called this range "borderline," a term that has aged poorly and implied far more than the data supports. In practice, scores in this band describe someone who finds the specific kinds of abstract, speeded reasoning that IQ tests sample harder than most people do — and say nothing about reliability, character, practical competence, or skill in domains the test never touches.
80–89 — Low Average (9th–25th percentile)
Roughly one person in six scores in this band. It sits entirely within the normal range of human variation; the "low" in the label is relative to the statistical center, nothing more. Someone at 85 is exactly as far from the mean as someone at 115 — one standard deviation — yet the two labels carry wildly different social charge, which says more about how we talk about intelligence than about the underlying statistics.
90–109 — Average (25th–75th percentile)
Half the population, by construction. If you score here, your abstract reasoning is typical of the people around you, which is the single most likely outcome for any person walking into any IQ test anywhere. The width of this band — twenty points — is a deliberate acknowledgment that scores wobble: two sittings by the same person can easily land ten points apart, and a band this wide absorbs most of that noise.
110–119 — High Average (75th–91st percentile)
Another one-in-six band, mirroring 80–89 on the other side of the mean. People here tend to find abstract and novel material noticeably easier than most, though in everyday life the difference between 105 and 115 is essentially invisible — far smaller than differences in knowledge, motivation, and work habits.
120–129 — Superior (91st–98th percentile)
About 7% of the population. This is the range where standardized-test facility becomes obvious — the person who found school exams straightforward. Studies of advanced-degree holders find their averages frequently in this vicinity, though with enormous overlap in every direction: plenty of people in this band never pursue academics, and plenty of accomplished academics score below it.
130–139 — Very Superior (98th–99.6th percentile)
Roughly 2% of people — about 1 in 44 — score 130 or above, which is why 130 became the conventional threshold for "gifted" programs and high-IQ societies. Two things are worth knowing at this altitude. First, the label is a convention, not a discovered boundary in nature; 129 and 131 describe essentially identical ability. Second, measurement precision starts to strain up here: there are fewer test items hard enough to separate ability levels, so scores get noisier exactly where people care most about exact numbers. Our guide to gifted-range scores covers what high scores do and don't predict.
140 and above — Exceptional (above the 99.6th percentile)
Fewer than 4 people in 1,000 score 140 or higher — around 1 in 260. At this extreme, ceiling effects dominate: most tests simply don't contain enough very hard, well-calibrated items to distinguish 140 from 150 reliably, and different tests disagree with each other most at exactly these heights. Any specific number above 140 should be read as "very far right tail" rather than as a precise measurement — including, and especially, from a 15-minute online test like ours.
Percentiles, not grades
The single most useful habit when reading an IQ score is to translate it immediately into a percentile, because percentiles say what the number actually claims. A score of 115 claims: "on this test, this person reasoned better than about 84% of the norming population." That's it. It isn't a grade out of 200, it isn't a measure of worth, and the scale isn't linear in any everyday sense — the fifteen points between 100 and 115 move you past 34% of the population, while the fifteen points between 130 and 145 move you past fewer than 2%, because so few people occupy the tail.
This is also why percentile comparisons feel so counterintuitive at the extremes, and why we report your position as "rarer than X% of people" on your AurorIQ result — the percentile framing is honest in a way the bare number never quite manages. For a deeper walkthrough, see reading your percentile.
Why exact numbers matter less than bands
Every psychological measurement carries a standard error, and IQ tests are no exception: on the best professionally administered instruments, the 95% confidence interval around a score spans roughly ±5 points; on shorter tests it's wider. Take a reasonably well-built test twice and your score can shift several points purely from chance, fatigue, time of day, or familiarity with the format — sleep and stress alone move scores measurably. Retaking the same test inflates scores further through practice effects.
The band and percentile range are therefore more honest representations of a result than the single number, which is why AurorIQ reports a confidence interval alongside every score, and why obsessing over a 3-point difference between two people — or two sittings — is reading precision into the instrument that the instrument does not possess.
Common misconceptions
"Your IQ is one fixed number." Scores are estimates with error bars, they drift with conditions, and childhood scores in particular are only moderately predictive of adult ones. What's relatively stable in adulthood is your broad position on the curve — not the third digit.
"Every IQ test gives the same score." Different tests use different content, different norming samples, and occasionally different scales entirely — the Cattell Culture Fair test uses a standard deviation of 24, so a Cattell 148 corresponds to roughly a Wechsler 130. Comparing raw numbers across tests without knowing the scale is meaningless.
"IQ is intelligence." IQ tests sample specific, measurable kinds of reasoning — the five domains AurorIQ measures are typical. They capture something real and predictive, but creativity, practical judgment, social insight, and accumulated expertise sit largely outside the instrument. A score describes performance on a particular class of problems, not the size of a mind. Our guide What is IQ, really? unpacks this at length.
"An online score means the same thing as a clinical score." It doesn't, and we'd rather tell you that than flatter you. A proctored assessment controls conditions, verifies identity, and uses instruments normed over decades. A good online adaptive test can produce a genuinely informative estimate — with a confidence interval — but it cannot substitute for clinical assessment, and no honest site claims otherwise. See our candid piece on how accurate IQ tests are.
Common questions
What is considered a good IQ score?
Scores between 90 and 109 fall in the average range, where roughly half the population sits. Scores above 110 are considered high average or higher. There's no single "good" score; the right framing is where it sits on the population curve, not a pass-or-fail line.
What is the average IQ?
100 — by definition rather than by measurement. IQ scales are periodically re-normed so that the population average always sits at 100, which means the "average IQ" can never drift even as raw performance changes across generations.
What percentile is an IQ of 130?
An IQ of 130 corresponds to roughly the 97th to 98th percentile, meaning a score that high or higher occurs in about 2 to 3 percent of the general population, assuming the standard normal distribution IQ scoring is built on.
Is an IQ of 120 good?
An IQ of 120 sits at roughly the 91st percentile — higher than about 9 in 10 people — at the entrance to the range traditionally labeled Superior. It's a genuinely high score, with the usual caveat that a few points either side of it describe essentially the same ability.
What IQ score do you need for Mensa?
Mensa admits the top 2% of the population, which corresponds to roughly 130–132 on tests with a standard deviation of 15, or 148 on the Cattell scale (standard deviation 24). Only supervised, approved tests qualify — online test scores, including AurorIQ's, are not accepted.
Is a 1-2 point IQ difference meaningful?
Generally no. Measurement error alone can shift a score by several points between sittings. The band a score falls into, and its percentile range, is far more informative than the exact number.
Can your IQ change over time?
Your relative position on the curve is fairly stable in adulthood, but individual scores can shift with health, sleep, stress, education, and measurement error. Childhood scores are notably less stable than adult ones. Large, permanent jumps in true ability are rare; moderate movement in measured scores is completely normal.
Why do different IQ tests give different scores?
Tests differ in content, norming samples, precision, and sometimes the scale itself — the Cattell test's standard deviation of 24 makes its numbers run higher than Wechsler-scaled tests for the same percentile. Add ordinary measurement error, and a 5–10 point spread between two well-built tests is unremarkable.