How this instrument works
A z-score answers one question: given where a value sits relative to its group's mean and spread, how unusual is it? A z-score of 0 means the value equals the mean exactly. A z-score of +1 means it sits one standard deviation above the mean; −2 means two standard deviations below. Because the formula divides by the standard deviation, z-scores are unitless — a z of +1.5 means the same thing whether the original values were exam scores, millimetres, or dollars, which is exactly what makes z-scores useful for comparing values that started out on completely different scales.
This is also called standardizing or normalizing a value. Once you convert several raw values from different distributions into z-scores, you can compare them directly: a student who scored 85 on a test with mean 75 (z = 1) and a runner who finished a race 1.5 standard deviations faster than average (z = −1.5, since faster is lower time) are both 'one-point-something standard deviations from typical' — a comparison that raw scores alone can't make, because a test and a stopwatch don't share units.
Under a roughly normal (bell-curve) distribution, z-scores also translate directly into percentiles: about 68% of values fall within z = ±1, about 95% within z = ±2, and about 99.7% within z = ±3 — the empirical rule. That's why a z-score beyond about ±2 or ±3 is often treated as flagging something genuinely unusual, not just ordinary noise.
- Enter the value you're checking into Raw score (x) — the actual measurement or result.
- Enter the group's average into Mean (μ) — the typical or expected value for that group.
- Enter the group's spread into Standard deviation (σ) — how much values typically vary from the mean; it must be greater than zero.
- Read Z-score — positive means above average, negative means below, and the magnitude tells you how many standard deviations away.
- To go the other direction — starting from a target z-score and finding the raw value it corresponds to — use this site's raw-score calculator instead.
Worked example — an 85 on a test with mean 75
A student scores 85 on a test where the class mean was 75 and the standard deviation was 10. Enter 85 into Raw score (x), 75 into Mean (μ), and 10 into Standard deviation (σ).
The instrument computes z = (85 − 75) / 10 = 1.0. The score sits exactly one standard deviation above the class average — under the empirical rule, that places the student above roughly the 84th percentile of the class, a clearly above-average but not extreme result.
Questions
What does a negative z-score mean?
It means the raw value is below the mean, not that anything is wrong with it. A z-score of −1.5 simply means the value sits one and a half standard deviations below average — for a test score that's below-average performance, but for a race time (where lower is faster) a negative z-score is actually a good result. The sign always just reflects direction relative to the mean; whether that direction is 'good' depends on what you're measuring.
How is a z-score different from a percentile?
A z-score measures distance from the mean in standard deviations; a percentile measures rank — the percentage of a distribution that falls below a given value. The two are connected under a normal distribution (z = 0 is the 50th percentile, z = 1 is roughly the 84th), but z-scores work directly with the raw scale of standard deviations, while percentiles require assuming or knowing the shape of the distribution to compute from a z-score.
What counts as an 'unusual' z-score?
There's no universal cutoff, but under the empirical rule for roughly normal data, about 95% of values fall within z = ±2 and about 99.7% within z = ±3. Many fields treat |z| > 2 as worth a second look and |z| > 3 as genuinely rare — but always check whether the underlying data is actually close to normally distributed, since the empirical rule doesn't hold for heavily skewed data.
Does the standard deviation have to be the population or sample value?
Either can be used — the formula is identical either way, since z-scoring just rescales by whatever spread figure you supply. What matters is being consistent: use the standard deviation that actually describes the group you're comparing the raw score against, whether that's a full population's σ or a sample's s.
Can I use this to compare scores from two different tests?
Yes — that's one of the main uses. Convert each raw score to a z-score using its own test's mean and standard deviation, then compare the z-scores directly. A z of 1.2 on Test A and a z of 0.8 on Test B tells you the Test A performance was relatively stronger, even if the two tests had completely different scoring scales.
I have a target z-score and want the raw value it corresponds to — is that the same tool?
No, that's the reverse calculation — this site's raw-score calculator does that, solving x = μ + z×σ for a target z instead of solving for z from a raw score. The two tools are exact inverses of each other: feed this calculator's answer into that one (or vice versa) and you'll get back your original input.