Z-Scores, Probabilities & Percentiles
Updated October 2026
The normal distribution — the familiar bell curve — describes how countless natural quantities cluster around an average: heights, test scores, measurement errors, and much more. This calculator takes a mean, a standard deviation, and a value, and tells you where that value sits: its z-score, the probability of landing below or above it, and its percentile. It turns an abstract curve into concrete answers about your specific number.
The reason the bell curve shows up so often is the Central Limit Theorem, which says that sums and averages of many independent influences tend toward a normal shape regardless of the underlying details. That's why it underpins so much of statistics, quality control, and the science of measurement.
A z-score answers a simple question: how many standard deviations is this value from the mean? A z of +2 means the value sits two standard deviations above average; a z of −1 means one below. Because the z-score strips out the original units, it lets you compare values from completely different distributions — a test score and a height percentile become directly comparable once both are z-scores. Our dedicated z-score calculator focuses just on that conversion.
One of the most useful facts about the normal distribution is the empirical rule: about 68% of values fall within one standard deviation of the mean, about 95% within two, and about 99.7% within three. So a value with a z-score beyond ±2 is genuinely unusual — it's in the outer 5% — and beyond ±3 is rare. This calculator shows the exact probabilities, but the rule gives you a quick mental benchmark.
The calculator reports the cumulative probability — the chance of a value falling below your input — which is the same as its percentile. The probability above is just one minus that. These figures are the foundation of statistical testing: a "p-value" is exactly this kind of tail probability. If you're building intervals around an estimate, the confidence interval calculator uses the same distribution.
Consider a standardized exam with a mean score of 500 and a standard deviation of 100, and suppose a student scores 650. Working out where that score falls shows exactly what the calculator does. The first step is always the same: convert the raw score into a z-score, which measures distance from the mean in standard-deviation units. Here, z = (650 − 500) ÷ 100 = 150 ÷ 100 = 1.5. The student scored one and a half standard deviations above average.
That z-score is the key that unlocks everything else, because probabilities for the standard normal distribution are fixed and tabulated. A z of 1.5 corresponds to a cumulative probability of about 0.9332. In plain terms, roughly 93.3% of test-takers scored at or below 650, placing this student at about the 93rd percentile. The probability of scoring higher is simply the complement, 1 − 0.9332 = 0.0668, or about 6.7%. So fewer than 1 in 14 students did better. None of this required knowing anything about the individual test questions — once you have the mean and standard deviation, the shape of the bell curve supplies the rest.
It is worth sanity-checking that result against the empirical rule. The rule says about 95% of values fall within two standard deviations of the mean, leaving 2.5% in each tail beyond ±2. A score of 650 sits at z = 1.5, comfortably inside two standard deviations, so a percentile in the low-to-mid 90s is exactly what you would expect — high, but short of the rarefied top 2.5% that a score of 700 (z = 2) would reach. This habit of cross-checking a precise figure against the 68-95-99.7 benchmark catches input errors quickly. Entering 500, 100, and 650 into the calculator returns this z-score, percentile, and tail probability at once, and the z-score calculator isolates that first conversion if that is all you need.
The normal distribution turns up so persistently that it can seem almost magical, but there is a concrete reason: the Central Limit Theorem. It states that when you add up or average many small, independent random influences, the result tends toward a normal shape regardless of how the individual influences are distributed. Human height is the classic illustration — it is the cumulative result of many genetic and environmental factors, no one of which dominates, and the population distribution comes out reliably bell-shaped. The same logic explains why measurement errors, manufacturing variations, and the averages of survey samples all cluster normally.
This is also what makes the distribution so useful in quality control and the sciences. A factory filling cereal boxes cannot make every box weigh exactly the same, but the weights will scatter normally around a target. By knowing the mean and standard deviation, an engineer can predict what fraction of boxes will fall outside an acceptable range and set the process accordingly — the foundation of statistical process control. In research, the normal distribution underlies the p-value: when a study reports statistical significance, it is comparing an observed result against how extreme that result would be if only normal random variation were at work.
A few cautions keep the tool honest. Not every dataset is normal — incomes, for instance, are famously skewed, with a long tail of high earners pulling the mean above the median, so applying bell-curve percentiles to income would mislead. Real data also has finite range, while the mathematical normal curve extends infinitely in both directions; for values many standard deviations out, the idealized probabilities are approximations. And the distribution is defined only for a positive standard deviation — a spread of zero would collapse the curve to a single spike, which has no meaningful percentiles. Used on data that genuinely clusters around a center, though, the normal distribution is one of the most powerful summarizing tools in all of statistics. For building a margin of error around an estimate, see the confidence interval calculator.
Use the z-score to compare across different scales — once two values are z-scores, a test result and a height become directly comparable.
Lean on the 68-95-99.7 rule for a sanity check: if your probability for a z of 2 isn’t near 95% within bounds, re-check your inputs.
The percentile and the "probability below" are the same number — don’t double-count them when interpreting a result.
A standard deviation of zero has no meaning here — the distribution would collapse to a single point, so the calculator needs a positive σ.