IQ scores follow a bell curve with a mean of 100 and a standard deviation of 15, and that single fact explains almost everything people misunderstand about intelligence testing. This guide walks through what the bell curve actually shows, what your score means in plain language, why exactly 68% of people land between 85 and 115, where the famous "genius" thresholds come from, and the limits of what an IQ score can tell you about a person.
Enter any IQ score and our tool plots you on the bell curve with your percentile, rarity (1 in N people), and standard deviations from the mean.
Open the IQ Bell Curve Calculator →What the IQ Bell Curve Is
An IQ score isn't a measurement of "how smart" you are in any absolute sense. It's a measurement of how you scored on a particular cognitive test relative to everyone else who took the same test. The bell curve — formally a normal distribution — is the shape that emerges when you plot how many people scored what.
The defining numbers, set by convention since the 1940s:
- Mean = 100. By definition. Whatever the average performance is on a properly standardized test, that score gets called 100. It's a label, not a measurement.
- Standard deviation = 15 on the Wechsler scales (the most widely used today). Some older tests used SD = 16 (Stanford-Binet) or 24 (Cattell). The number you see compared to "100" only makes sense if you know which SD scale was used.
From those two numbers, the entire bell-curve geometry follows.
What Your Score Actually Means
Because IQ tests are designed to produce a normal distribution, you can convert any score directly to a percentile rank. Here's the map:
| Score range | Percentile | Population share | Common label |
|---|---|---|---|
| ≥ 145 | ≥ 99.9 | ~0.1% | "Highly gifted" |
| 130–144 | 98–99.8 | ~2% | "Gifted" |
| 115–129 | 84–97 | ~14% | "Above average" |
| 85–114 | 16–84 | ~68% | "Average range" |
| 70–84 | 2–15 | ~14% | "Below average" |
| ≤ 69 | ≤ 2 | ~2% | "Significantly below average" |
Two practical implications most people get wrong:
"Average" is a huge band, not a single number. Anyone who scores between 85 and 114 sits in the middle 68% of the population. The difference between IQ 90 and IQ 110 is real, but it's smaller than people imagine — both are well within the range where individual variation in skill, motivation, and education matters far more than the IQ gap.
The tails matter for very specific things. Once you're above ~130 or below ~70, the score starts predicting concrete outcomes (gifted-program admission, eligibility for certain disability supports). In the middle 95% of the distribution, IQ alone is a weak predictor of most life outcomes that people care about.
Why Exactly 68%, 95%, and 99.7%?
The three percentages come from a mathematical property of the normal distribution called the 68-95-99.7 rule:
- 68% of scores fall within ±1 standard deviation of the mean (IQ 85–115).
- 95% of scores fall within ±2 standard deviations (IQ 70–130).
- 99.7% of scores fall within ±3 standard deviations (IQ 55–145).
This isn't a feature of human intelligence — it's a feature of the test design. IQ tests are standardized, meaning their scoring is deliberately calibrated to produce a normal distribution on a representative sample. The bell curve isn't discovered. It's enforced. If a test produced a skewed or bimodal distribution, the psychometricians would rescale it until the bell appeared.
That's worth pausing on: the bell-curve shape doesn't prove anything about the nature of intelligence. It only proves that the test was successfully calibrated against a reference population.
What "Genius" Actually Means
Popular culture treats IQ 140 as the threshold for "genius," but there's no official cutoff. The boundaries that do have institutional meaning:
- IQ ≥ 130 (top 2%) — the standard threshold for "gifted" programs in most school systems and for organizations like Mensa.
- IQ ≥ 145 (top 0.1%) — sometimes called "highly gifted" or "exceptionally gifted." About 1 in 1,000 people.
- IQ ≥ 160 (top 0.003%) — "profoundly gifted." About 1 in 30,000. Single-test measurement at this level has very high error bars because there aren't enough test items hard enough to discriminate among people this far out.
Above roughly IQ 150, the tests stop being reliable for ranking individuals. Two people who both score 165 may genuinely differ in cognitive ability by more than the test can detect, because the test wasn't designed to discriminate at that extreme.
The Score You See Is an Estimate, Not a Reading
This is the single most important fact about your own IQ score, and most people miss it. Every IQ test produces a confidence interval, typically ±5 points on professionally administered Wechsler tests. If your tested score is 117, your "true" IQ — the score you'd average over hundreds of testing sessions — is almost certainly between 112 and 122.
The implication: small differences in IQ scores are noise, not signal. A friend who tested at 124 and you who tested at 119 don't have measurably different cognitive ability. Take any specific score with a grain of salt; take the percentile range around it more seriously.
Online IQ tests, which are taken without a proctor and often without proper norming, have wider error bars still — often ±10 points or more. They're useful for entertainment and broad orientation, but not for any high-stakes decision. If you want to see your own ballpark, a free online IQ test like ours gives you a directional read; for a clinically valid score, you need a licensed psychologist administering the WAIS-IV or similar in person.
What the Bell Curve Predicts — and What It Doesn't
Decades of research have measured what IQ correlates with. The honest summary:
Where IQ predicts something useful: academic performance, the speed at which someone learns complex new tasks, performance on cognitively demanding jobs (engineering, medicine, law), and certain test-administered outcomes. The correlations are real but moderate — typically in the 0.3–0.5 range, meaning IQ explains maybe 10–25% of the variance in these outcomes.
Where IQ predicts surprisingly little: career satisfaction, relationship quality, happiness, leadership effectiveness, creative achievement past a moderate threshold, ethical behavior, and almost everything that requires sustained effort over years. Past about IQ 120, additional IQ stops correlating much with real-world success because other factors — conscientiousness, social skill, opportunity, persistence — dominate.
The bell-curve number is one input. It tells you something. It doesn't tell you the most important thing about a person, and it doesn't tell you most of what predicts how their life will go.
The Bell Curve Book (Brief, Honest Note)
The phrase "bell curve" became politically loaded after Herrnstein and Murray's 1994 book The Bell Curve, which argued that group IQ differences had social-policy implications. The book's empirical claims about test scores are largely accurate; its interpretations and policy recommendations have been widely contested by other psychologists. If you've heard the bell curve discussed in a controversial frame, that book is usually why. The underlying psychometrics — the mean-100, SD-15 normal distribution described above — predates the book by half a century and isn't itself controversial.
Frequently Asked Questions
What is a normal IQ score?
Any score between 85 and 115 is in the "average range" and covers roughly two-thirds (68%) of the population. The single most common score is 100, by definition.
What percentile is an IQ of 120?
Approximately the 91st percentile. About 9 in 100 people score higher.
What's the rarest IQ score?
Scores above 145 (top 0.1%) and below 55 (bottom 0.1%) are statistically the rarest. Anything above 160 is so rare that single-test measurement becomes unreliable.
Is the bell curve shape natural or designed?
Designed. IQ tests are deliberately calibrated to produce a normal distribution. The bell curve reflects the test design, not a direct property of human intelligence itself.
How accurate is my IQ score?
Even professionally administered tests have a confidence interval of about ±5 points. Online tests have wider error bars, often ±10 points or more. Treat any specific score as an estimate within a range, not a precise measurement.