What IQ scores mean
An IQ score is a rank dressed up as a number. It says where you sit relative to everyone else who took the same test, and nothing at all about how much of anything you have.
The scale is defined, not measured
The modern IQ scale is a deviation scale: the population average is fixed at 100 and one standard deviation is fixed at 15. Those two numbers are chosen, not discovered. A test does not find out that the average person scores 100; the scoring is built so that they do.
This means a score only exists relative to a reference population. The same performance can produce different numbers on two tests normed on different groups, which is why comparing a score from one test to a score from another is usually meaningless.
What each band covers
Because the scale is normal, scores bunch in the middle. Roughly two thirds of people fall between 85 and 115, and about 95% fall between 70 and 130.
About 2 in 100 people.
Two standard deviations above the mean. Above roughly 145 a fixed-length test has too few hard items left to separate people, so the number stops meaning much.
About 14 in 100 people.
One to two standard deviations up. This is the band where the gap between adjacent scores is smallest in practice, because most items discriminate best here.
About 68 in 100 people.
The middle two thirds of the population sit inside this 30-point band. A 96 and a 108 are not meaningfully different people.
About 14 in 100 people.
One to two standard deviations down.
About 2 in 100 people.
Interpreting scores here needs a supervised assessment, not an online one. The precision of any fixed-length test collapses at the tails.
The same scale as percentiles
A percentile is easier to reason about than a deviation score, because it states the rank directly. These are the conversions this test uses, computed from the normal distribution:
The full table covers every score from 55 to 145.
Why a single number is not enough
Every fixed-length test has measurement error, and that error is not constant across the scale. Mid-range scores are estimated from many items of roughly the right difficulty, so they are precise. Extreme scores are estimated from the handful of items hard or easy enough to be informative, so they are not.
That is why a result should be reported as an interval. A score of 118 with a 95% interval of 112 to 124 is an honest statement: the best estimate is 118, and a repeat sitting would plausibly land anywhere in that range. A score reported as a bare number is hiding this.
How this test estimates both the score and its interval is set out in the methodology.
What the number does not cover
The scale indexes a specific, narrow thing: performance on abstract reasoning problems under time pressure, without domain knowledge. It does not index judgement, motivation, creativity, expertise, or the ability to hold a job or a friendship. Those are not being measured, so the score is silent on them.