Coefficient of Variation Calculator
Enter the mean and the standard deviation to get the coefficient of variation — the single percentage that says how variable a dataset is relative to its average.
Two numbers in, one percentage out
Enter the mean and the standard deviation and the calculator returns the coefficient of variation, CV% = (σ ÷ μ) × 100.
The mean cannot be zero
The CV divides by the mean, so a mean of zero leaves it undefined. Use it for ratio data with a true zero and a non-zero average.
What is the coefficient of variation?
Standard deviation as a percentage of the mean
The coefficient of variation (CV), sometimes called the relative standard deviation, expresses the standard deviation as a percentage of the mean. Instead of reporting spread in raw units, it reports spread relative to the average — so a CV of 10% means the standard deviation is one tenth of the mean. Because the units cancel out, the CV is a unitless measure of relative spread: you can compare the variability of datasets that use completely different units or scales, something a raw standard deviation can never do. That makes it the go-to figure for judging consistency, whether you are comparing the steadiness of lab measurements, the volatility of returns, or the uniformity of manufactured parts.
Enter the mean and the standard deviation to get the coefficient of variation instantly.
One short formula: divide the standard deviation by the mean, then turn it into a percentage.
CV% = (σ ÷ μ) × 100Here the Greek letter σ (sigma) is the standard deviation and μ (mu) is the mean. Dividing σ by μ gives the spread as a fraction of the average, and multiplying by 100 turns that fraction into a percentage. The mean must not be zero, otherwise the formula would divide by zero and the CV would be undefined. The result carries no units, which is exactly what lets you compare it across datasets.
Suppose a dataset has a mean of 50 and a standard deviation of 5.
Note the mean and the spread
The mean is 50 and the standard deviation is 5 — both on the same scale.
Divide the spread by the mean
5 ÷ 50 = 0.1 — the standard deviation is one tenth of the mean.
Convert to a percentage
0.1 × 100 = 10% — the coefficient of variation is 10%.
The single CV percentage carries a lot of meaning once you know how to read it. Because the CV strips out the original units, it is a unitless measure of relative spread — its whole purpose is to let you compare variability across datasets with different units or scales. A lower CV means the data is more consistent relative to its average: a CV of 5% describes far steadier data than a CV of 40%, even if the second dataset happens to have a smaller raw standard deviation. As a rough guide, a CV under about 10% is often treated as low variability, 10–30% as moderate, and above 30% as high — but these bands are conventions, not laws, and a sensible threshold depends entirely on your field. The CV is most meaningful for ratio data with a true zero, such as prices, weights, distances, or counts, where the mean is naturally positive and a percentage of it makes intuitive sense. It is undefined when the mean is zero and becomes unstable when the mean is very close to zero, since tiny changes in a near-zero average swing the percentage wildly. For data that can be negative or that lacks a true zero, such as temperatures in Celsius, the CV can mislead, so reach for it only when a percentage of the mean is genuinely informative.
The formula is exact, but a few practical points are worth keeping in mind.
A non-zero mean, ratio data, and a sensible benchmark
The coefficient of variation only makes sense when the mean is non-zero and the data sits on a ratio scale with a true zero — prices, weights, or counts, not temperatures in Celsius or arbitrary scores. It is undefined at a mean of zero and unstable when the mean is near zero, since small shifts in the average then swing the percentage dramatically. There is no universal "good" CV either: always judge it against typical values for your own type of data rather than a fixed cut-off, and pair it with the raw standard deviation and mean so you never lose sight of the original scale.