Skip to main content
ponderloop

Z-Score Calculator

A z-score says how many standard deviations a value sits from the mean: z = (x − μ) ÷ σ. For a value of 130 with a mean of 100 and a standard deviation of 15, z = 2. On a normal curve that is the 97.7th percentile; the left-tail probability (percentile) is 0.977250 (97.725%), the right tail is 0.022750 (2.275%) and the two-tailed probability is 0.045500 (4.55%). Enter your own value, mean and standard deviation, or paste a list of numbers and let the calculator find the mean and the sample or population standard deviation, or go the other way and turn a percentile into a z-score and a value. The normal curve below shades the area you choose.

Updated Runs in your browser: nothing you enter is uploadedHow this tool works

z = (x − μ) ÷ σ = (130 − 100) ÷ 15

z = 2

130 is 2 standard deviations above the mean: the 97.7th percentile, with 97.725% of a normal distribution below it and 2.275% above.

Z-score
2
Value (x)
130
Mean (μ)
100
Standard deviation (σ)
15
Left tail, P(Z ≤ z): the percentile
0.977250 (97.725%)
Right tail, P(Z ≥ z)
0.022750 (2.275%)
Two-tailed, P(|Z| ≥ |z|)
0.045500 (4.55%)
Middle, P(|Z| ≤ |z|)
0.954500 (95.45%)

Normal curve

Standard normal curve with the shaded areaThe standard normal curve from −4 to 4 with the left tail, P(Z ≤ 2) shaded, an area of 97.725%. The z-score is 2.−3−2−10123z = 2
Shaded: the left tail, P(Z ≤ 2), 97.725% of the area under the curve.

Z-scores and the areas under the standard normal curve

Each row is computed by the calculator’s own normal distribution, to 4 decimals as in a printed z-table.
zLeft tail P(Z ≤ z)Right tail P(Z ≥ z)Two-tailed P(|Z| ≥ |z|)Middle P(|Z| ≤ |z|)
00.50000.50001.00000.0000
0.50.69150.30850.61710.3829
10.84130.15870.31730.6827
1.50.93320.06680.13360.8664
1.6450.95000.05000.10000.9000
1.960.97500.02500.05000.9500
20.97720.02280.04550.9545
2.5760.99500.00500.01000.9900
30.99870.00130.00270.9973

Critical z-values for common confidence levels

The z that leaves the stated tail area beyond it, from the inverse of the normal CDF.
Confidence levelTwo-tailed z*One-tailed z*
80%1.28160.8416
90%1.64491.2816
95%1.961.6449
98%2.32632.0537
99%2.57582.3263
99.9%3.29053.0902

Percentile to z-score

The z-score at each percentile of a normal distribution.
Percentilez-score
1st−2.3263
5th−1.6449
10th−1.2816
25th−0.6745
50th0
75th0.6745
90th1.2816
95th1.6449
99th2.3263

The 68–95–99.7 rule

The share of a normal distribution within a number of standard deviations of the mean.
WithinShare of values
±1 standard deviation (z from −1 to 1)68.2689%
±2 standard deviations (z from −2 to 2)95.45%
±3 standard deviations (z from −3 to 3)99.73%

How to calculate a z-score

  1. Find the mean and standard deviation. Use the mean (μ) and standard deviation (σ) of the group: 100 and 15 in the example. For a list of numbers, paste them in and the calculator works both out, using n − 1 for a sample or n for a whole population.
  2. Subtract the mean from the value. 130 − 100 = 30. The sign says which side of the mean the value is on.
  3. Divide by the standard deviation. 30 ÷ 15 = 2. The value is 2 standard deviations above the mean.
  4. Read the probability. The standard normal CDF gives the area to the left of z: 0.977250 (97.725%) for z = 2. One minus that is the right tail (0.022750 (2.275%)); twice the smaller tail is the two-tailed probability (0.045500 (4.55%)).

Frequently asked questions

What is a z-score?
A z-score (or standard score) is the number of standard deviations a value is above or below the mean: z = (x − μ) ÷ σ. A z of 0 is the mean itself, +1 is one standard deviation above it and −1 is one below. In the example, 130 with mean 100 and standard deviation 15 gives z = 2.
How do I find the percentile from a z-score?
The percentile is the area under the standard normal curve to the left of z, which is the left-tail probability Φ(z) times 100. For z = 2 that is 0.977250 (97.725%), so the value is at about the 97.7th percentile. This assumes the data are roughly normally distributed; for other shapes the z-score still measures distance in standard deviations, but the percentile can differ.
What is the difference between left-tail, right-tail and two-tailed probability?
The left tail P(Z ≤ z) is the area below z, the right tail P(Z ≥ z) is the area above it, and the two-tailed probability P(|Z| ≥ |z|) adds both ends, the area at least as far from the mean as z in either direction. For z = 2: left 0.977250 (97.725%), right 0.022750 (2.275%), two-tailed 0.045500 (4.55%), and the middle area between −2 and 2 is 0.954500 (95.45%). A two-tailed p-value is twice the smaller one-tailed value.
How do I find the value (x) for a given percentile?
Switch to “A percentile”, enter the percentile and the mean and standard deviation. The calculator finds z with the inverse normal CDF and then x = μ + zσ. For the 95th percentile with mean 100 and standard deviation 15, z is ≈ 1.644854 and x is ≈ 124.672804. With no mean and standard deviation you get the plain z-score: the 97.5th percentile is z ≈ 1.959964.
What is a negative z-score?
A negative z-score means the value is below the mean. For example, 85 with mean 100 and standard deviation 15 has z = −1, a left-tail probability of 0.158655 (15.8655%). The curve is symmetric, so a z-score of −z has the same tail areas as +z with the tails swapped.
Which z-scores count as unusual?
There is no single cut-off, but under a normal distribution about 95.45% of values lie within 2 standard deviations of the mean, so |z| above 2 happens about 4.55% of the time, and |z| above 3 about 0.27% of the time. Many people flag |z| > 2 as unusual and |z| > 3 as an outlier, but that depends on the data and on how many values you are checking.
Should I use the sample or the population standard deviation?
Use the population standard deviation (divide by n) when your list is every value in the group, and the sample standard deviation (divide by n − 1) when it is a sample from a larger group. The calculator uses the standard deviation calculator’s code for both. For the list 2, 4, 4, 4, 5, 5, 7, 9 the mean is 5; scoring 9 gives z = 2 with the population standard deviation (2).
How is the probability calculated, and how accurate is it?
The standard normal CDF is built from the error function, which the calculator evaluates with W. J. Cody’s rational Chebyshev approximations (accurate to about 15 digits), and the right tail and two-tailed values are computed directly so that tiny tail probabilities keep their precision. The tool’s values are tested against published normal tables (the NIST/SEMATECH e-Handbook and Abramowitz and Stegun) to 7 decimals, and shown to 6 decimals. The inverse uses Peter Acklam’s approximation refined with a Halley step, so a percentile round-trips to z and back.
Why is the standard deviation required to be above 0?
A standard deviation of 0 means every value is identical, so there is no spread to measure against and z = (x − μ) ÷ 0 is undefined. The calculator says so instead of showing a made-up number, both for a typed standard deviation and for a list of equal numbers.
Z-score or t-score: which should I use?
A z-score assumes you know the population standard deviation, or have a large sample. When you only have a small sample and estimate the standard deviation from it, tests and confidence intervals use the t distribution, whose tails are heavier. The z-score itself, the distance from the mean in standard deviations, is calculated the same way either way.
How is the z-score of a value from a pasted list calculated?
Numbers are read by the same parser as the average and standard deviation calculators (commas, spaces and new lines separate them; text that is not a number is ignored and listed). The mean and the chosen standard deviation are found exactly, and your value x is scored against them. The list is only used on your device.
What is the 95th percentile as a z-score?
The 95th percentile is z = 1.6449: 95% of a normal distribution lies below it. The familiar 1.96 is the 97.5th percentile, which is where a middle 95% interval ends on each side.

How it works

The z-score is z = (x − μ) ÷ σ. For a list of numbers, the mean and the sample (n − 1) or population (n) variance are computed exactly with the standard deviation calculator’s code and then converted to a standard deviation. Probabilities use the standard normal distribution: Φ(z) = ½ erfc(−z ÷ √2), with the error function and its complement evaluated by W. J. Cody’s rational Chebyshev approximations (Math. Comp. 23, 1969), which stay accurate in the far tails. The right tail is ½ erfc(z ÷ √2), the two-tailed probability is erfc(|z| ÷ √2) and the middle area is erf(|z| ÷ √2), each computed directly rather than as 1 − p. The inverse (percentile to z) uses Peter Acklam’s rational approximation followed by a Halley step against the accurate Φ. In the test suite, Φ is checked against the NIST/SEMATECH e-Handbook’s published table of the standard normal CDF and against an independent series and continued fraction for the error function. Probabilities are shown to 6 decimals, or in exponent form below 0.000001; z-scores and values to 6 decimals with ≈ when rounded. The percentile must be strictly between 0 and 100. The curve is drawn from −4 to 4, and a z-score beyond that is marked at the edge. The normal model is an assumption: z-scores are exact distances in standard deviations for any data, but the probabilities only describe data that are roughly normal. Everything runs in your browser and nothing you type is sent anywhere.

Sources

Guides

More: Statistics, probability and gradesAll math & numbersAll toolsHome & DIY calculators