The Standard (Normal) Distribution of IQ Scores: Mean, SD, and Percentiles
People searching for the standard normal distribution of IQ scores are usually trying to turn a score into an understandable position on a bell curve. The familiar reference scale places the mean at 100 and commonly uses a standard deviation of 15. That convention lets us describe how far a score is from the norm-group average and estimate a percentile, but it does not turn a test result into a complete description of a person.
The most useful way to read the curve is as a statistical map: first identify the test and its norm group, then use its mean, standard deviation, and published norm table. The model is helpful for explaining patterns such as “about two thirds within one standard deviation,” while real scores still include measurement error, rounding, and departures from perfect normality.
What is a standard normal distribution?
The normal distribution is a symmetric, single-peaked probability model described by a mean (μ) and standard deviation (σ). The standard normal distribution is the special version with μ = 0 and σ = 1. A raw observation can be converted to that common scale with a z-score:
z = (score − mean) ÷ standard deviation
For a deviation-IQ scale with mean 100 and standard deviation 15, the reverse conversion is approximately:
IQ = 100 + 15z
Thus, an IQ of 115 is one standard deviation above the norm mean (z = +1), while 85 is one standard deviation below it (z = −1). Some instruments use a different standard deviation, such as 16, so the formula must match the test manual rather than being assumed from a headline chart.
Ready to discover your IQ?
Take our scientifically designed test and get your score in just a few minutes.
What does the 68–95–99.7 rule mean for IQ?
For an ideal normal curve, the empirical rule gives the approximate share of observations in each band. Applying mean 100 and SD 15 produces this useful illustration:
| Distance from the mean | Approximate share | IQ interval on a 100/15 scale |
|---|---|---|
| Within ±1 SD | 68.27% | 85–115 |
| Within ±2 SD | 95.45% | 70–130 |
| Within ±3 SD | 99.73% | 55–145 |
The percentages describe areas under a model. They are not a promise that exactly 68.27% of every school, website, or country sample will fall between 85 and 115. Sampling, selection, age composition, language, and test conditions can all change the observed distribution.
How do z-scores become IQ percentiles?
A z-score expresses distance in standard-deviation units; a percentile expresses the proportion of a reference population scoring at or below a point. Under the standard normal model, z = 0 is the 50th percentile, z = +1 is about the 84th percentile, and z = +2 is about the 97.7th percentile. The corresponding IQ examples on a 100/15 scale are:
| IQ | Approximate z-score | Approximate percentile* |
|---|---|---|
| 85 | −1.00 | 16th |
| 100 | 0.00 | 50th |
| 115 | +1.00 | 84th |
| 130 | +2.00 | 97.7th |
| 145 | +3.00 | 99.9th |
*These are model-based approximations. A test’s official norm table, score rounding, and age-specific reference group take priority. Percentile is not the percentage of questions answered correctly: a 97th-percentile score does not mean 97% of items were correct.
Are mean, median, and mode the same on an IQ bell curve?
In a perfectly symmetric normal distribution, the mean, median, and mode coincide. On a norm-referenced IQ scale, test developers often transform scores so that the norming sample is centered near 100, which is why 100 is close to the 50th percentile. In a finite sample, however, rounding and random variation can make the three summaries differ slightly.
The distinction becomes important outside the norming sample. A gifted-program group may have a ceiling effect and a crowded high end. A clinical referral sample may be concentrated below the norm mean. An online quiz can attract people who are unusually interested in testing. In each case, reporting only “average IQ” hides the shape that determines whether a mean is representative.
Ready to discover your IQ?
Take our scientifically designed test and get your score in just a few minutes.
What does norming have to do with the curve?
Norming is the process of administering a test to a defined standardization group and converting raw performance into reference scores. The American Psychological Association describes a norm as a typical standard based on a large group, and a norm-referenced score is interpreted against that population. The relevant group may be divided by age, language, country, or other characteristics.
That is why an IQ score is not meaningful without the instrument and norm edition. A published percentile can change when a test is renormed, even if the raw number of items a person got right does not change. Continuous-norming methods can model age more finely than broad age bands and may improve percentile estimates, but they still depend on representative data and appropriate modeling choices.
Is every real IQ sample normally distributed?
No. A normal curve is a reference model, not a diagnostic label for every dataset. Actual scores can be skewed by selection, truncated by floor or ceiling limits, or show more than one peak when distinct subgroups are mixed. Online samples and convenience samples are especially poor substitutes for a test’s carefully recruited norm group.
Before applying a z-score interpretation to a group, inspect its sample size, histogram, missing data, score range, and recruitment method. A statistical test may also assess normality, but visual inspection and subject-matter context matter because very large samples can flag trivial departures and small samples can miss important ones. If the data are not approximately normal, report medians, quantiles, or an empirical percentile table alongside the mean and standard deviation.
How should an individual interpret a score near the tail?
Tail scores are sensitive to small changes. Moving from IQ 129 to 130 can cross a commonly discussed threshold, yet the difference may be smaller than the test’s measurement error. A confidence interval gives a range of plausible true scores; it is more honest than treating a single observed number as exact. Retesting can also introduce practice effects, and different instruments may not be interchangeable.
Use the score report’s standard error, confidence interval, composite definition, and age norm. A Full-Scale IQ is a composite, not a direct measurement of every cognitive ability. Fluid reasoning, working memory, processing speed, language, health, sleep, anxiety, and testing conditions can each affect performance. The bell curve helps locate a result in a reference distribution; it does not establish fixed potential or explain why two people received their scores.
What are the limits of converting IQ to a percentile?
Percentile conversion is most defensible when the score comes from a valid administration and the person matches the norm group. Extrapolating a percentile beyond the test’s observed range can create false precision, especially at very high or low scores where only a small number of norm participants are available. Different tests can use different SDs, composite rules, and smoothing procedures, so their percentiles should not be blended casually.
For a practical report, name the test and edition, age and language norms, score type, standard deviation, percentile or confidence interval, and the date of interpretation. If you are summarizing a group, provide the sample size and distribution rather than presenting the normal model as a ranking of people or countries.
Q: What is the normal distribution of IQ scores?
A: It is a bell-shaped reference model centered near 100 on many modern IQ scales. With SD 15, the model places about 68% of scores between 85 and 115 and about 95% between 70 and 130.
Q: What percentage of people have an IQ between 85 and 115?
A: The ideal 100/15 normal model predicts about 68.27%. The exact percentage in a real group depends on its sampling, norm group, and whether the observed scores follow the model.
Q: What percentile is an IQ of 130?
A: It is approximately the 97.7th percentile under a mean-100, SD-15 normal model. Use the test’s official norm table for the reported percentile, because rounding and instrument-specific norms can change the result.
Q: Does an IQ score have to be normally distributed?
A: No. The bell curve is often used to construct or explain norm-referenced scores, but a particular sample can be skewed, truncated, or multimodal. Check the actual data before making a normality assumption.
Q: Does the normal distribution prove that IQ is fixed or biological destiny?
A: No. It describes the spread of scores in a reference population, not the causes of an individual result. Measurement error, learning, health, context, and test conditions all matter, and a score should be interpreted with its confidence interval.
References
- American Psychological Association. IQ and standardized score.
- National Institute of Standards and Technology. Normal distribution.
- American Psychological Association. Test norm.
- Sijtsma, K., et al. Continuous norming and age-related score interpretation.
Last updated: July 19, 2026
✨Related Articles
Which Myers-Briggs (MBTI) Type Has the Highest IQ? What Research Shows
No MBTI type has a scientifically established highest IQ. MBTI describes preferences, while intelligence tests measure selected cognitive abilities on a separate scale.
The Lowest IQ Ever Recorded: Why There Is No Reliable World Record
Is there an official lowest IQ ever recorded? Learn why IQ-test floors, measurement error, adaptive behavior, and outdated claims make a single world-record number impossible to verify.
IQ Scale by Age for Women: How Female Scores Should Be Interpreted
A modern IQ score is age-normed, so 100 is the scale mean for women and men in the relevant reference group. Research shows domain-specific and age-related patterns, not a fixed female IQ ladder.