📉 Standard Normal Z-Score & Percentile Calculator
A z score calculator both ways — score to percentile and back — with the step most skip: for a sample mean the divisor is σ/√n, not σ.
Critical values (percentile → z)
| Percentile / confidence level | z |
|---|---|
| 90th percentile | 1.2816 |
| 95th percentile (one-tailed 95%) | 1.6449 |
| 97.5th percentile (two-tailed 95%) | 1.9600 |
| 99th percentile | 2.3263 |
| 99.5th percentile (two-tailed 99%) | 2.5758 |
What Standard Normal Z-Score & Percentile Calculator Does
A z-score answers one question: how many standard deviations away from the mean is this? Subtract the mean, divide by the standard deviation, and the units cancel — which is the whole point. A z of +1.5 means the same thing whether the underlying data was measured in kilograms, dollars or exam marks, and that is what makes scores from different scales comparable.
There is a second version of the formula that trips people up, and the top-ranking calculator for this term glosses over it. When you are standardizing a single observation, you divide by the standard deviation. When you are standardizing a *sample mean* — asking how unusual an average of n observations is — you divide by the standard error, which is the standard deviation divided by the square root of n.
The difference is not cosmetic. Take a population with a mean of 100 and a standard deviation of 15. A single value of 105 gives z = 0.33, entirely unremarkable. An average of 105 across 100 people gives z = 3.33, which would happen by chance about four times in ten thousand. Same gap from the mean, completely different conclusion, because averages vary far less than individuals do.
Once you have a z, converting it to a percentile means finding the area under the standard normal curve to its left. That is what a z-table does, and what this page does in both directions — score to percentile, and percentile back to the score.
How to Use Standard Normal Z-Score & Percentile Calculator
- Enter the raw value, the population mean and the standard deviation
- Leave sample size at 1 for a single observation, or set it to n to standardize a sample mean
- Read the z-score, its percentile, and the one- and two-tailed p-values
- Use the critical-value table to go the other way, from a percentile or confidence level back to z
Formula Used by Standard Normal Z-Score & Percentile Calculator
Standardizing a single value
z = (x − μ) ÷ σ
- x
- The raw value you want to standardize
- μ
- The population mean
- σ
- The population standard deviation
Worked example
A test score of 85 where the mean is 70 and the standard deviation is 10
- 85 − 70 = 15
- 15 ÷ 10
Result: z = 1.50 — the score sits one and a half standard deviations above the mean, at about the 93.3rd percentile
Standardizing a sample mean (the step most calculators skip)
z = (x̄ − μ) ÷ (σ ÷ √n)
- x̄
- The mean of your sample
- σ ÷ √n
- The standard error of the mean — how much sample averages of size n vary
- n
- How many observations went into that average
Worked example
μ = 100, σ = 15, and a sample mean of 105
- n = 1: standard error 15 ÷ √1 = 15, so z = 5 ÷ 15 = 0.33
- n = 9: standard error 15 ÷ 3 = 5, so z = 5 ÷ 5 = 1.00
- n = 25: standard error 15 ÷ 5 = 3, so z = 5 ÷ 3 = 1.67
- n = 100: standard error 15 ÷ 10 = 1.5, so z = 5 ÷ 1.5 = 3.33
Result: The same 5-point gap goes from unremarkable to extreme purely because of sample size
z to percentile
The percentile is the area under the standard normal curve to the left of z — the proportion of the distribution falling below that score.
| z | Percentile | Area to the right |
|---|---|---|
| −3.00 | 0.13% | 99.87% |
| −2.00 | 2.28% | 97.72% |
| −1.00 | 15.87% | 84.13% |
| 0.00 | 50.00% | 50.00% |
| +1.00 | 84.13% | 15.87% |
| +1.645 | 95.00% | 5.00% |
| +1.96 | 97.50% | 2.50% |
| +2.00 | 97.72% | 2.28% |
| +2.576 | 99.50% | 0.50% |
| +3.00 | 99.87% | 0.13% |
Percentile to z — the critical values worth memorizing
The inverse direction. Note that a "95% confidence" z depends on whether the test is one-tailed or two-tailed, which is where 1.645 and 1.96 come from.
| You want | z |
|---|---|
| 90th percentile | 1.2816 |
| 95th percentile — one-tailed 95% | 1.6449 |
| 97.5th percentile — two-tailed 95% | 1.9600 |
| 99th percentile | 2.3263 |
| 99.5th percentile — two-tailed 99% | 2.5758 |
| 25th percentile (lower quartile) | −0.6745 |
| 10th percentile | −1.2816 |
Why sample size changes everything
A population with mean 100 and standard deviation 15, against an observed value or average of 105.
| n | Standard error σ/√n | z | Percentile |
|---|---|---|---|
| 1 (a single value) | 15.000 | 0.3333 | 63.06% |
| 9 | 5.000 | 1.0000 | 84.13% |
| 25 | 3.000 | 1.6667 | 95.22% |
| 100 | 1.500 | 3.3333 | 99.96% |
How to Read Your Result
"How many z-scores is 95%?" has two answers
It depends on whether you care about one direction or both. If you only care about values being unusually high, 95% of the distribution sits below z = 1.6449 — that is the one-tailed critical value. If you care about being unusually far from the mean in either direction, you need 2.5% in each tail, which puts the boundaries at z = ±1.96. The 1.96 figure is the one people half-remember as "about two standard deviations", and it is where the familiar 95% confidence interval comes from.
The sample-size trap
This is the single most common error with z-scores, and the top-ranking calculator for this search does not warn about it. Its treatment of the sample case says only that the sample mean is used in place of the population mean, with no mention of the square root of n. If you are asking "is this one measurement unusual?", divide by σ. If you are asking "is this average unusual?", divide by σ/√n. Getting it wrong understates your z by a factor of √n — a tenfold error at n = 100.
z-tables are not interchangeable
There are several conventions in circulation, and picking the wrong one silently shifts every answer. Some tables give the cumulative area from the far left up to z. Others give the area between the mean and z, which is the cumulative value minus 0.5. Others give only the right tail. Calculator.net uses the mean-to-z form: their table shows 0.36864 for z = 1.12, whereas a cumulative table would show 0.86864 for the same score. Check which kind you are reading before trusting a number off it.
What a z-score assumes
Converting a z into a percentile assumes the underlying data is approximately normally distributed. The z-score itself is always computable — it is just arithmetic — but the percentile attached to it is only meaningful if the bell-curve assumption roughly holds. For strongly skewed data, income being the classic case, a z of 2 does not correspond to the 97.7th percentile and treating it as though it does will mislead you badly.
z or t?
The z-score assumes you know the population standard deviation. In practice you usually do not — you have estimated it from the same sample you are testing, which adds uncertainty the normal distribution does not account for. The t-distribution exists for exactly that case and has slightly fatter tails. The two converge as the sample grows and are nearly identical past about n = 30, which is why the rough convention of using z for large samples and t for small ones persists.
Limitations & Accuracy Notes
- The percentile conversion assumes an approximately normal distribution. The z-score is valid arithmetic for any data, but the probability attached to it is not.
- This tool uses the population standard deviation. If you estimated the standard deviation from your own sample, a t-distribution is technically more appropriate, particularly below about 30 observations.
- The normal cumulative function is computed by a standard numerical approximation accurate to roughly seven decimal places. That is far beyond the precision of any real dataset, but it is an approximation rather than an exact closed form, which does not exist.
- Clinical z-scores — paediatric growth, cardiac dimensions such as aortic root or coronary artery measurements — use population-specific regression models, not this general formula. Several specialized tools exist for those, and they are not interchangeable with this one.
- A z-score describes where a value sits in a distribution. It says nothing about whether the value is good, whether the difference is important, or whether anything caused it.
Frequently Asked Questions
What is a z-score?
How many z-scores is 95%?
What is the z-score for the 90th percentile?
Do I divide by the standard deviation or the standard error?
Why does my z-table give a different number?
What does a z-score actually tell me?
What is the 68-95-99.7 rule?
Does a z-score require a normal distribution?
Is a z-score of 3 an outlier?
Can a z-score be negative?
Is my data stored?
References & Further Reading
- NIST/SEMATECH e-Handbook — Normal distribution — The standard normal distribution, its cumulative function, and the properties the percentile conversion depends on
- NIST/SEMATECH e-Handbook — Critical values of the normal distribution — Reference table behind the one- and two-tailed critical values quoted above