Empirical Rule Calculator

Apply the 68-95-99.7 rule to any normal distribution — enter mean and standard deviation to get the three ±σ ranges with a shaded bell-curve diagram.

The empirical rule explained, step by step

The empirical rule — also called the 68-95-99.7 rule — is a quick approximation for how values spread in an approximately normal (bell-shaped, symmetric) distribution. Each card below covers one of the three σ-bands the calculator reports, plus the assumption that makes the rule work.

1. ~68% within ±1σ — the middle of the bell

About 68% of values in a roughly normal distribution sit within one standard deviation of the mean. Shift the mean and the band moves with it; change σ and the band widens or narrows.

P(μ − σ ≤ X ≤ μ + σ) ≈ 0.6827
μmean·σstandard deviation
Example
Given
μ = 75, σ
= 8
Substitute
75 ± 8
Answer
[67, 83]

2. ~95% within ±2σ — the working confidence band

Adding a second σ on each side captures about 95% of the distribution. This is why 2σ appears everywhere in applied statistics — from control-chart limits to the informal '95% confidence' shortcut.

P(μ − 2σ ≤ X ≤ μ + 2σ) ≈ 0.9545
Example
Given
μ = 75, σ
= 8
Substitute
75 ± 16
Answer
[59, 91]

3. ~99.7% within ±3σ — 'almost everything'

Three sigmas on each side cover roughly 99.7% of the distribution — only about 0.3% of values fall outside. This is where the 'three-sigma rule' in quality control comes from, and why 3σ is a common outlier flag for normal data.

P(μ − 3σ ≤ X ≤ μ + 3σ) ≈ 0.9973
Example
Given
μ = 75, σ
= 8
Substitute
75 ± 24
Answer
[51, 99]

4. When the rule breaks — skew, bimodality, heavy tails

The 68-95-99.7 percentages come from a normal curve. If the data is skewed, bimodal, heavy-tailed or discrete with few values, the σ-bands stop matching those percentages. When you can't assume normality, Chebyshev's inequality (at least 1 − 1/k² inside k σ) still holds for any distribution.

Chebyshev: P(|X − μ| ≤ kσ) ≥ 1 − 1/k²
skewed → percentages shift
bimodal → no single bell fits
heavy tails → 3σ too optimistic
Example
Given
k = 2 (any distribution)
Substitute
1 − 1/4
Answer
≥ 75%

Features of this calculator

  • Instant μ ± 1σ, μ ± 2σ and μ ± 3σ bounds with their approximate ~68% / ~95% / ~99.7% shares.
  • Shaded bell-curve diagram with the three bands in progressively lighter shades, labeled with the actual numeric boundaries from your inputs.
  • Optional X input: enter a specific value to see which band it falls into and its approximate percentile from the normal CDF.
  • Show/hide step-by-step working laying out μ ± kσ arithmetically for k = 1, 2, 3.

Frequently asked questions

+Why the specific numbers 68, 95 and 99.7?

They're the actual areas under a standard normal curve between ±1, ±2 and ±3 standard deviations, rounded to convenient values. The exact percentages are 68.27%, 95.45% and 99.73%.

+How do I know if my data is 'normal enough' for the rule?

Plot a histogram and look for a roughly symmetric bell shape. For a more formal check, a Q-Q plot or a normality test (Shapiro-Wilk, Anderson-Darling) is standard. Mild deviations are usually fine for a quick empirical-rule estimate.

+What's outside ±3σ called?

Values beyond ±3σ are commonly flagged as outliers in normal populations because only about 0.3% of the distribution sits out there. In quality control this is the origin of the "three-sigma control limit".

+Can I use the rule on sample data?

Yes, using the sample mean x̄ and sample standard deviation s as estimates of μ and σ. Just remember the rule is an approximation and small samples produce noisy s values, so the bands may shift as you collect more data.

+What percentage is between 1σ and 2σ?

About (95% − 68%) / 2 ≈ 13.5% on each side of the mean. So roughly 13.5% of values lie between μ + 1σ and μ + 2σ, and another 13.5% between μ − 2σ and μ − 1σ.

Related calculators