Average IQ Score Explained
- Responsible organization
- MOSAIC-48
- First published
- Last reviewed
- Review standard
- Facts, limitations, and score explanations are reviewed against the published method and primary references.
The short answer is 100 — but the number is more interesting than it looks. IQ scales are engineered so that the average is always 100, with a standard deviation of 15. That is a design choice, not a fact of nature, and understanding how the scale is built is the key to reading any score — including your own — without over- or under-reading it.
Below we walk through how the 100 ± 15 scale works, what each score band means, how percentiles map onto it, and the traps to avoid when you interpret a single result.
How the 100 ± 15 scale works
IQ scores use a normal distribution — the familiar bell curve — centred on a mean of 100 with a standard deviation (SD) of 15. Raw performance on the underlying tasks is transformed so that the group average lands exactly on 100 and the spread matches an SD of 15. This is why 100 is always average: the scale is defined that way, not discovered.
The standard deviation is the unit of the scale. One SD up is 115, two is 130, three is 145; one SD down is 85, and so on. Because the curve is symmetric and well understood, those steps translate directly into how common a score is:
- About 68% of people score within one SD — between 85 and 115.
- About 95% score within two SDs — between 70 and 130.
- Only about 2% score above 130, and about 2% below 70.
Score bands and the percentiles they map to
Because the scale is a comparison, every score corresponds to a percentile — the share of the reference group scoring at or below it. The bands below are rounded for readability, so read the percentiles as approximate:
- 130 and above — very high range — roughly the top 2% (about the 98th percentile and up).
- 120 to 129 — high range — roughly the 91st to 97th percentile (about the top 3–9%).
- 110 to 119 — above average — roughly the 75th to 90th percentile.
- 90 to 109 — the average range where most people fall — roughly the 25th to 73rd percentile.
- 80 to 89 — below average — roughly the 9th to 23rd percentile.
- 79 and below — low range — roughly the bottom 8% (below the 9th percentile).
Why "average" is reset over time
Because 100 is defined as the average rather than discovered, it is not tied to any fixed level of raw performance. Across the twentieth century, raw scores on cognitive tasks drifted steadily upward from one generation to the next — a well-documented pattern researchers call the Flynn effect. If nothing were done about it, the typical person today would appear to score well above 100 against an older reference group.
To keep the centre at 100, tests are periodically re-standardized against a fresh, representative sample. Each re-norming pulls the average back to 100, so a score always means "relative to a recent reference group," never "relative to some absolute, timeless standard." It is one more reason a raw number in isolation says so little: it only takes on meaning against the norm it was scaled against.
Reading a score with care
Notice how wide the middle band is: a 20-point span from 90 to 109 covers almost half of all people, a useful reminder that scores near 100 are barely distinguishable in practice. A single number also hides a margin of error. Every score carries measurement error, so the same person can land a few points apart on two sittings, and sleep, stress, practice with the format, and motivation all nudge the result.
Two habits keep you honest. First, read a score as a range, not a point — "somewhere in the low hundreds" is more truthful than "exactly 104." Second, remember that an average of several abilities can look identical for two very different people: the same composite can hide sharp strengths and weaknesses that only a domain profile reveals.
Try a few example questions
Two quick questions on the percentile idea behind the bands above.
Question 1 of 3
In a group of 200 people, about 2% score above a certain cutoff. Roughly how many people is that?
About MOSAIC-48 scores
MOSAIC-48 places its composite near a familiar 100-centred range so the number is easy to read at a glance, but that shared look does not make it the same scale. The score bands and percentiles above describe standardized IQ tests, and they do not convert a MOSAIC-48 composite: a MOSAIC figure is only an estimate against an internal reference distribution assumed for scoring, not a placement on the standardized 100/15 norm those bands are built on. It is not an official IQ test and does not apply standardized clinical norms — its figures are internal-reference estimates meant for self-insight, not certified scores, and they cannot be used for diagnosis, selection, or any medical decision.
Where MOSAIC-48 tries to add value is the shape, not the single number: alongside the composite it shows a profile across eight domains, so two people with the same overall figure can still see how differently they arrived there.
Frequently asked questions
What is a "good" IQ score?
By construction, 100 is exactly average, and most people fall between 85 and 115. Rather than chasing a threshold, it is more useful to look at your profile of strengths across different abilities.
Is a score of 100 bad?
No — 100 is the exact middle of the scale. Roughly half of people score below it and half above, so 100 is the very definition of typical, not a weak result.
Can my score change over time?
Day-to-day conditions and ordinary measurement error move any single result by several points. Look for a stable pattern across attempts rather than treating one number as fixed.
Has the average IQ changed over time?
The average is always reset to 100 by re-standardizing the test, so by definition it stays at 100. What has changed is raw performance: across the twentieth century it rose from one generation to the next, which is why tests are periodically re-normed against a fresh reference group.
See your own cognitive profile
MOSAIC-48 is a full online composite cognitive assessment across eight domains. It is not an official IQ test and does not apply standardized clinical norms — it reports internal-reference estimates for self-insight, not a diagnosis.