What Your IQ Score Means: Percentiles, Ranges and Retakes

Your IQ score tells you where your performance sits relative to a reference group, usually people your age. On the standard scale, 100 is average and 15 points is one standard deviation, so 115 means you outscored about 84% of the group and 85 means you outscored about 16%. A percentile states that comparison directly. Either way, the score is an estimate with a margin of error, not an exact measurement of a fixed quantity.

Percentile is not percent correct

The most common misreading goes like this: "I'm at the 70th percentile, so I got 70% right." A percentile rank says how many people in the reference group scored below you. It says nothing directly about how many questions you answered correctly. On a hard test, getting 55% of items right could place you at the 90th percentile; on an easy one, 90% correct might be merely average.

Percentiles also depend entirely on who is in the reference group. A 70th percentile on two different tests means the same standing only if both tests compare you with similar groups of people.

Converting a percentile to an IQ, and back

On a test normed against the general population, IQ and percentile are linked by the normal distribution. The table shows common points on both scales, rounded to the nearest whole IQ point.

Two things stand out. First, the scales are not linear. Near the middle, a few IQ points cover many people: moving from the 50th to the 60th percentile takes about 4 IQ points, while moving from the 89th to the 99th takes roughly 16. Second, the top percentiles look crowded even when the underlying gaps are large, which is why "99th percentile" covers everyone from about 135 upward. The full scale is in our IQ table from 55 to 145.

One caution: this conversion is only valid when the percentile comes from a representative norm group. Our own test measures your position against the people who have taken it (here and on My Path, where the same test runs), not against the general population. Its result page does convert that position with this same math and shows it as an estimated IQ, always with a range and labelled as an estimate — so read that number as "where I stand among this test's takers", never as a population IQ. Why that distinction matters is explained in how IQ scores are calculated.

Your score is a range, not a point

Every test score contains measurement error: luck on particular items, fatigue, attention, noise in the room. Psychometricians quantify it with the standard error of measurement (SEM), which depends on how reliable the test is:

SEM = standard deviation × √(1 − reliability)

Major clinical batteries report full-scale reliability coefficients in the mid-to-high .90s in their technical manuals. At a reliability of .97, the SEM is about 2.6 points, so a 95% confidence interval spans roughly ±5 points. For a hypothetical shorter scale with a reliability of .80, the SEM grows to about 6.7 points and the 95% interval to roughly ±13.

So a reported IQ of 108 from a good clinical test really means "probably somewhere between about 103 and 113" (manuals center the interval slightly toward 100, so it is not perfectly symmetric)." Clinicians report the interval for exactly this reason, and the professional Standards for Educational and Psychological Testing expect publishers to document measurement error. Any online test should show it too.

Our test shows its result as a range of ±12 percentile points on the 12-question short form and ±7 on the 60-question full form. Those are deliberately conservative fixed widths rather than a measured standard error, because the sample is still too small to estimate one well. The longer form gets the narrower range because more items average out more of the noise.

When your domain scores disagree

Most modern tests report several domain or index scores alongside the overall number, such as verbal comprehension, working memory and processing speed. People are often surprised by how much these differ. Wechsler test manuals include base-rate tables precisely because sizable gaps between index scores, 15 points or more between two of them, turn up regularly in their normative samples among people with no clinical condition at all.

A few guidelines for reading that scatter:

• Small gaps are mostly noise. Each domain score rests on fewer items than the total, so it is less reliable. Differences that fall inside overlapping confidence intervals tell you little.
• Large, repeatable gaps can be informative. A big difference that shows up across testing occasions can reflect a real relative strength or weakness. Interpreting it is a clinician's job, and a gap on its own does not indicate a learning disability, ADHD or anything else.
• What the overall score means under heavy scatter is debated. Some clinicians argue that a single number summarizing very uneven abilities says less about the person; others argue it often remains a useful summary. Both camps agree the profile should be reported, not hidden.

On a short online form, be especially cautious. Our short form has two items per domain, one easier and one harder, which is far too few to support any conclusion about a specific strength. The full form has ten per domain: better, but still a sketch rather than a profile.

What your score does and does not tell you

Across large groups, IQ scores predict real outcomes. Strenze's 2007 meta-analysis of longitudinal studies found correlations of about .56 with later educational attainment, .43 with occupational level and .20 with income. Those are meaningful associations, but they leave most of the variation between individual lives unexplained. Your score says nothing about your motivation, character, practical skills or what you will do with your abilities.

The bands people use to describe scores are summarized in IQ score ranges. The labels attached to them differ between tests and editions: the WAIS-IV called scores of 130 and above "Very Superior," while the WISC-V and the WAIS-5 use "Extremely High" for the same range. A label is a convention, not a finding.

If you need a score for a decision, such as an evaluation for intellectual disability, a learning disability or a gifted program, only an assessment by a licensed psychologist counts. Nothing on this page is medical or educational advice.

Should you retake an IQ test?

Retaking a test usually raises your score whether or not your ability has changed. A 2007 meta-analysis by Hausknecht and colleagues found average gains of roughly a quarter of a standard deviation on a second attempt at a cognitive ability test, about 4 IQ points. A later meta-analysis by Scharfen, Peters and Holling (2018) found that retest gains grew over the first few repetitions before leveling off.

Some practical rules follow:

• Clinical tests: examiners generally avoid repeating the same instrument at short intervals, commonly waiting a year or more, because practice effects persist. When re-evaluation is needed sooner, they may use a different test.
• Online tests: retaking the same items mostly measures your memory of them. If you took a short form, the more informative next step is a longer form, not the same short form again.
• Extreme results: a very high or very low first score will tend to land closer to average next time, because part of any extreme result is luck. This is regression to the mean, not evidence that you got smarter or less smart.
• Bad days: if you were exhausted, ill or interrupted, a retake after rest is reasonable. Waiting weeks rather than minutes reduces the memory advantage.

Real change in measured ability over years is possible, especially with schooling; our article on whether IQ can change over time covers what the evidence shows.