Z-score calculator

A z-score turns any measurement into the same unit: how many standard deviations it sits from the mean.

What a z-score does

A z-score restates a value as its distance from the mean, measured in standard deviations:

z = (x − μ) / σ

An IQ of 130 with mean 100 and standard deviation 15 gives z = 2. That means 2 standard deviations above average, which sits at roughly the 97.7th percentile.

The point of standardising is comparison. A score of 85 on one test and 42 on another are not comparable as raw numbers, but if the first is z = 0.5 and the second is z = 1.8, the second performance was clearly stronger relative to its cohort.

Reading the number

z = 0 — exactly average.
Positive — above the mean. Negative — below it.
|z| < 1 — inside the middle 68%. Unremarkable.
|z| > 2 — outside the middle 95%. Worth noticing.
|z| > 3 — outside 99.7%. Rare, and in real data often a sign of a measurement error rather than a genuine extreme.

Those thresholds come from the empirical rule: for a normal distribution, about 68% of values fall within 1 standard deviation of the mean, 95% within 2, and 99.7% within 3.

Percentiles depend on normality

The z-score itself is just arithmetic and always valid. The percentile is not — converting z to a percentile assumes the underlying data follow a normal distribution.

Plenty of real data does not. Income is heavily right-skewed, so a z-score of 2 on income does not correspond to the 97.7th percentile at all. Reaction times, waiting times and anything bounded at zero tend to be skewed too. Check the shape of your distribution before trusting the percentile; the z-score can still be reported honestly as "two standard deviations above the mean".

Population versus sample

This calculator uses μ and σ — population parameters. If you are working from a sample, you would normally use the sample mean and sample standard deviation (with n−1 in the denominator), and for small samples the t-distribution is more appropriate than the normal.

Where n is large, say above 30, the difference becomes negligible and the normal approximation is fine. Below that, using z where t is required understates how uncertain you actually are.

Common uses

Standardised test reporting, quality control limits set at ±3σ, outlier detection in cleaning datasets, and grading on a curve all rest on z-scores. In hypothesis testing, the critical values you memorise are z-scores: 1.96 for a two-tailed test at 5%, and 2.576 at 1%.

Common questions

What does a z-score of 2 mean?

The value sits two standard deviations above the mean. If the data are normally distributed that is roughly the 97.7th percentile, so about 2.3% of values are higher.

Can a z-score be negative?

Yes. A negative z-score simply means the value is below the mean. The sign carries direction and the magnitude carries distance, so -2 and +2 are equally far out.

Do I need normally distributed data?

For the z-score itself, no — it is just arithmetic. For the percentile, yes. Converting z to a percentile assumes normality, and on skewed data such as income that conversion is misleading.

What counts as an outlier?

A common convention is a z-score beyond plus or minus 3, which covers only 0.3% of a normal distribution. Some fields use 2.5. In practice, an extreme z is often worth checking as a data-entry error before treating it as real.