Empirical Rule Calculator
Find the ranges containing approximately 68%, 95%, and 99.7% of an approximately normal dataset—and see exactly where a value falls.
For approximately bell-shaped data
Enter the distribution mean and a positive standard deviation. You can also enter an observation to calculate its z-score and approximate percentile.
95% Empirical-Rule Range
—
Within 1σ
≈ 68%—
—
About 16% remains in each tail.
Within 2σ
≈ 95%—
—
About 2.5% remains in each tail.
Within 3σ
≈ 99.7%—
—
About 0.15% remains in each tail.
Normal Distribution Map
The colored bands show distance from the mean.
Checked Value
—
—
How to Use This Calculator
- Enter the mean, or center, of the distribution.
- Enter a standard deviation greater than zero.
- Optionally enter one observation to see its z-score and approximate percentile.
- Select the desired rounding, then calculate the three empirical-rule ranges.
- Use the result only when the data are reasonably symmetric and bell-shaped.
What Is the Empirical Rule?
An empirical rule calculator estimates how much normally distributed data falls within 1, 2, or 3 standard deviations of the mean. The empirical rule states that about 68% of values lie within 1 standard deviation, 95% within 2, and 99.7% within 3 standard deviations.
68%
μ − σ to μ + σ
95%
μ − 2σ to μ + 2σ
99.7%
μ − 3σ to μ + 3σ
How to Read Every Band and Tail
The 68–95–99.7 totals can be split into one-sided regions. This makes it possible to answer questions such as “What percentage is above two standard deviations?” without starting a new calculation.
| Region | Approx. share | Why |
|---|---|---|
| Below μ − 3σ | 0.15% | Half of the 0.3% outside the three-sigma range |
| μ − 3σ to μ − 2σ | 2.35% | Half of 99.7% − 95% |
| μ − 2σ to μ − 1σ | 13.5% | Half of 95% − 68% |
| μ − 1σ to μ | 34% | Half of the central 68% |
| μ to μ + 1σ | 34% | The normal curve is symmetric around μ |
| μ + 1σ to μ + 2σ | 13.5% | Half of 95% − 68% |
| μ + 2σ to μ + 3σ | 2.35% | Half of 99.7% − 95% |
| Above μ + 3σ | 0.15% | Half of the 0.3% outside the three-sigma range |
Above μ + 1σ
≈ 16%
Above μ + 2σ
≈ 2.5%
Above μ + 3σ
≈ 0.15%
The same percentages apply below the matching negative boundary. For example, approximately 84% lies below μ + 1σ and 97.5% lies below μ + 2σ.
Normal-curve reference: UC Berkeley — The Normal Distribution.
Empirical Rule Formulas
Lower endpoint
Lower = μ − kσ
Upper endpoint
Upper = μ + kσ
Use k = 1 for the 68% range, k = 2 for the 95% range, and k = 3 for the 99.7% range. The total width of each interval is 2kσ, so changing the mean shifts every endpoint while changing the standard deviation expands or contracts the ranges.
Worked Example: Test Scores
Suppose a bell-shaped set of scores has a mean of 80 and a standard deviation of 6. The 68% range is 74 to 86, the 95% range is 68 to 92, and the 99.7% range is 62 to 98. A score of 92 has z = 2, placing it two standard deviations above the mean and near the 97.7th percentile under an exact normal model.
When the Rule Is—and Is Not—Appropriate
Reasonable use
- The histogram is roughly symmetric and bell-shaped.
- There are no dominant gaps, extra peaks, or extreme outliers.
- You need quick, approximate ranges rather than an exact probability.
Use another method
- The data are strongly skewed, bounded, or multimodal.
- The distribution is unknown and the sample is too small to assess its shape.
- You need the probability for an arbitrary boundary rather than ±1σ, ±2σ, or ±3σ.
Check Whether the Empirical Rule Fits Your Data
A mean and standard deviation are enough to perform the arithmetic, but they do not prove that the underlying distribution is normal. Use the raw data when available.
Inspect a histogram
Look for one central peak, similar left and right sides, and tails that taper gradually. Strong skew, separated clusters, or hard floor and ceiling effects weaken the normal approximation.
Compare center measures
The mean and median should be reasonably close for symmetric data. A large gap can signal skewness, although agreement by itself does not establish normality.
Use a normal Q–Q plot
Points that follow an approximately straight line support a normal model. Systematic curves, tail departures, or isolated points reveal skewness, unusually heavy or light tails, and possible outliers.
Compare observed counts
Count the observations inside ±1σ, ±2σ, and ±3σ and compare them with 68%, 95%, and 99.7%. Allow for ordinary sampling variation, especially in smaller datasets.
If the shape does not fit
Report the median, quartiles, percentiles, or the observed proportion in a range instead of forcing the empirical rule. Investigate unusual values before excluding them, and use a distribution-specific model only when its assumptions are defensible.
Diagnostic reference: University of Virginia Library — Understanding Q–Q Plots.
Empirical Rule vs. Other Ways to Describe Spread
Choose the method that matches what is known about the distribution and how precise the answer must be.
| Method | Assumption | Within 2σ | Best use |
|---|---|---|---|
| Empirical rule | Approximately normal or bell-shaped | About 95% | Fast mental estimate at 1σ, 2σ, or 3σ |
| Exact normal calculation | A normal probability model is appropriate | About 95.45% | Precise probabilities, percentiles, and arbitrary cutoffs |
| Chebyshev’s theorem | Any distribution with finite mean and variance | At least 75% | A guaranteed lower bound when shape is unknown |
| Observed proportion | Representative raw data are available | Count it directly | Describing what occurred in the actual dataset |
Quick approximation
Use the empirical rule when the shape is credible and the boundary is exactly 1σ, 2σ, or 3σ from the mean.
Exact probability
Convert the raw value with z = (x − μ) ÷ σ, then use the standard normal cumulative distribution.
Unknown shape
For k > 1, Chebyshev guarantees at least 1 − 1/k² of values within k standard deviations.
Distribution-free reference: Encyclopedia of Mathematics — Chebyshev inequality.
Frequently Asked Questions
Is the empirical rule exact?
No. The 68 percent rule, 95 percent rule, and 99.7 percent rule are rounded descriptions of the central region under a normal distribution. Exact probability-density areas are about 68.27% within one standard deviation, 95.45% within two standard deviations, and 99.73% within three standard deviations. The calculator uses the familiar rounded percentages for the ranges and an exact normal approximation for the optional z-score result.
What is the difference between the empirical rule and Chebyshev’s theorem?
The empirical rule gives tighter percentages when normality and symmetry make a bell curve a reasonable model. Chebyshev’s theorem works with any probability distribution having a finite mean and variance, so its guaranteed proportions are more conservative: at least 75% within two sigma and at least 88.9% within three sigma.
Does a value beyond three standard deviations count as an outlier?
It is an unusual value under a normal model because only about 0.3% of observations lie outside the ±3σ interval—approximately 0.15% in the left tail and 0.15% in the right tail. Treat that distance from the center as a reason to investigate, not automatic proof of an error; a valid measurement or a non-normal distribution can also produce an extreme observation.
Can I use a sample mean and sample standard deviation?
Yes, when the sample is representative and the continuous variable is reasonably bell-shaped. The arithmetic mean, also called the average, is a sample statistic that estimates the population’s center; the sample standard deviation estimates its dispersion. Report the dataset size because both parameter estimates can be unstable when the sample is small or contains influential values.
What happens when the standard deviation changes?
Standard deviation measures spread in the original units, while variance is its square. Increasing the deviation moves each lower bound and upper bound farther from the mean: the interval width is 2σ for the one-sigma range, 4σ for the two-sigma range, and 6σ for the three-sigma range. The covered percentages do not change because they are defined by standardized distance.
Is a 95% empirical range a confidence interval?
No. The μ ± 2σ range is part of descriptive statistics and estimates where individual data values occur. A confidence interval instead quantifies uncertainty in an unknown population parameter, such as the mean, using the sampling distribution and standard error. The two intervals can both contain 95% in their labels while answering different questions.
What do z-score, z-value, and standard score mean?
They are different names for the same statistic: z = (value − mean) ÷ standard deviation. The sign identifies the side of the curve, and the absolute z-value gives the distance from the mean in standard-deviation units. Under a normal model, the area to the left gives the approximate percentile and the remaining area gives the probability above that value.
How can I check whether my dataset is approximately normal?
Start with a histogram or frequency plot. Look for one peak, gradual tails, and similar shape on both sides of the center; the mean and median should also be reasonably close. A normal Q–Q plot offers a stronger visual check: points near a straight line support the model, while curves or separated points suggest skewness, heavy tails, or outliers. These checks support an estimate of normality but do not prove that the data follow a perfect normal curve.
Other Useful Calculators
Elimination Method Calculator
Solve two-variable linear systems step by step.
Generic Rectangle Calculator
Calculate common rectangle measurements.
Polynomial Division Calculator
Find a polynomial quotient and remainder.
Complex Root Calculator
Calculate roots involving complex numbers.
Normal Force Calculator
Find the normal force in a mechanics problem.
Resultant Velocity Calculator
Combine components to find resultant motion.