Calculate the percentage from a sigma value or find sigma from a standard normal cumulative probability using the inverse CDF formula for z-scores.
Sigma to Percentage Formula
The following formula is used to convert a sigma level to a percentage:
Variables:
- P is the left-tail cumulative percentage under a standard normal model
- z is the standard normal score (no 1.5 shift)
For the default left-tail mode, evaluate the standard normal CDF at z and multiply by 100. Right-tail mode uses 100 × (1 − Φ(z)); central mode uses 100 × (2Φ(z) − 1) for nonnegative z. Select the corresponding inverse mode to recover z from a percentage.
| Sigma (z) | Percentage (%) |
|---|---|
| -3.00 | 0.135 |
| -2.50 | 0.621 |
| -2.00 | 2.275 |
| -1.96 | 2.500 |
| -1.645 | 5.000 |
| -1.50 | 6.681 |
| -1.00 | 15.866 |
| -0.6745 | 25.000 |
| -0.50 | 30.854 |
| 0 | 50.000 |
| 0.50 | 69.146 |
| 0.6745 | 75.000 |
| 1.00 | 84.134 |
| 1.2816 | 90.000 |
| 1.50 | 93.319 |
| 1.645 | 95.000 |
| 1.96 | 97.500 |
| 2.00 | 97.725 |
| 2.50 | 99.379 |
| 3.00 | 99.865 |
| * Percentages are left-tail probabilities Φ(z) × 100 for the standard normal distribution, rounded to 3 decimals. | |
What is Sigma Level?
Sigma level is a statistical measure derived from the standard normal distribution that quantifies how far a data point or process mean sits from a specification limit, expressed in standard deviations. In pure statistics, a sigma value (z-score) maps directly to the cumulative distribution function (CDF) of the normal distribution, giving the probability that a random observation falls below that threshold. A z-score of 1.96 corresponds to the 97.5th percentile, which is the basis for the 95% confidence interval used across clinical trials, A/B testing, and social science research.
In quality management, sigma level describes process capability. A process operating at 3 sigma produces 66,807 defects per million opportunities (DPMO), equating to a 93.3% yield. At 6 sigma, DPMO drops to 3.4 with a 99.99966% yield. The distinction matters because the statistical z-score and the Six Sigma process level use the same scale but apply it differently: the z-score converts directly through the normal CDF, while Six Sigma conventions incorporate a 1.5-sigma shift to account for long-term process drift.
Sigma Level, DPMO, and Yield Reference Data
The relationship between sigma level, defects per million opportunities, and process yield follows a precise mathematical progression through the normal distribution. Defect rates fall nonlinearly with sigma. In the shifted convention, moving from 3 to 4 reduces DPMO by about 10.8 times (66,807 to 6,210); moving from 4 to 5 reduces it by about 26.7 times (6,210 to 233). The table below includes the 1.5-sigma long-term shift standard in Six Sigma methodology.
| Sigma Level | DPMO | Yield (%) | Defect Rate (%) |
|---|---|---|---|
| 1 | 691,462 | 30.85 | 69.15 |
| 2 | 308,538 | 69.15 | 30.85 |
| 3 | 66,807 | 93.32 | 6.68 |
| 4 | 6,210 | 99.38 | 0.62 |
| 5 | 233 | 99.977 | 0.023 |
| 6 | 3.4 | 99.99966 | 0.00034 |
The 1.5 Sigma Shift
The +1.5 sigma adjustment is a common Six Sigma conversion convention, not a universal measured drift of every process. Under that convention, a reported level of 6 maps to an unshifted one-sided Z of 4.5 and about 3.4 DPMO. An unshifted Z of 6 has about 0.000987 DPMO in one tail, or 0.001973 DPMO across both tails.
When using the calculator above, the conversion uses the standard normal CDF without the 1.5 shift. To estimate long-term process performance in a Six Sigma context, subtract 1.5 from your short-term sigma level before looking up the corresponding percentage.
Industry Sigma Benchmarks
Sigma levels vary dramatically across industries based on process complexity, regulatory requirements, and acceptable risk thresholds. These benchmarks provide context for interpreting your own sigma calculations.
No universal industry sigma ranges are supplied: rates depend on the dataset, time period, defect/opportunity definitions and convention. Compare documented like-for-like data.
Sigma in Statistical Hypothesis Testing
For a standard normal variable, the central area within ±2 is about 95.45%, and within ±3 is about 99.73%. These central areas differ from cumulative left-tail percentages. A statistical test threshold and its interpretation depend on whether the test is one- or two-sided and on the model assumptions; there is no universal minimum sigma for all research.
Under a specified standard-normal null model, the two-sided tail areas beyond ±2, ±3 and ±5 are about 4.55%, 0.27% and 0.0000573%, respectively. A p-value describes how incompatible the data are with that model; it is not the probability that an observed result happened by chance or that the hypothesis is true.
