Convert between the three ways Six Sigma describes process quality. Enter defects, units and opportunities to get DPO, DPMO, yield and the sigma level (using the standard 1.5σ long-term shift plus the short-term Z), or work backward from a target sigma level to the DPMO and yield it implies.
| Sigma | DPMO | Yield |
|---|---|---|
| 1σ | 691,462 | 30.85% |
| 2σ | 308,537 | 69.15% |
| 3σ | 66,807 | 93.32% |
| 4σ | 6,210 | 99.379% |
| 5σ | 233 | 99.9767% |
| 6σ | 3.4 | 99.99966% |
Six Sigma measures process quality in three interchangeable ways: DPMO (defects per million opportunities), the sigma level, and yield (the fraction of defect-free output). This converter moves freely between them. Enter raw defect counts to get DPMO, yield, and the sigma level — or start from a target sigma level and see the DPMO and yield it requires. It uses the industry-standard 1.5σ long-term shift and also reports the short-term Z-score.
Start with three counts. Defects (D) is the number of defects found. Units (U) is the number of items inspected. Opportunities per unit (O) is the number of distinct ways each unit could be defective — counting opportunities, not just whole-unit pass/fail, makes the metric fair across products of different complexity.
Defects Per Opportunity: DPO = D / (U × O). Defects Per Million Opportunities: DPMO = DPO × 1,000,000 — the headline Six Sigma number. Yield (first-pass, at the opportunity level) = (1 − DPO) × 100%. Example: 15 defects across 1,000 units with 3 opportunities each gives DPO = 15 / 3,000 = 0.005, DPMO = 5,000, yield = 99.5%.
The sigma level expresses how many process standard deviations fit between the process mean and the nearest specification limit — higher is better. Six Sigma's defining claim is that a "6 sigma" process produces just 3.4 DPMO. That figure builds in a 1.5σ long-term shift: the idea that over time a process mean drifts by about 1.5 standard deviations, so the long-term defect rate corresponds to a short-term capability 1.5σ better.
This tool uses the Schmidt/Launsby approximation to get the (shifted, long-term) sigma level directly from DPMO: σ ≈ 0.8406 + √(29.37 − 2.221·ln(DPMO)). It is accurate to about ±0.02σ across the practical range and avoids needing a Z-table.
Two related numbers describe the same process. The short-term Z-score is the true number of standard deviations to the spec limit right now, computed directly from the defect rate: Z = Φ⁻¹(1 − DPO), where Φ⁻¹ is the inverse normal. The long-term sigma level adds the 1.5σ shift to account for drift: sigma level = short-term Z + 1.5.
That is why a process running at 3.4 DPMO is called "6 sigma" (long-term) even though its short-term Z is about 4.5. This converter shows both so you are never confused about which convention a number follows — always confirm whether a quoted sigma figure includes the shift.
With the 1.5σ shift: 2σ ≈ 308,537 DPMO (about 69% yield), 3σ ≈ 66,807 DPMO (93.3% yield — a common starting point for many uncontrolled processes), 4σ ≈ 6,210 DPMO (99.38%), 5σ ≈ 233 DPMO (99.977%), and 6σ ≈ 3.4 DPMO (99.99966%).
Notice the dramatic, non-linear payoff: moving from 3σ to 4σ cuts defects more than tenfold, and 4σ to 5σ cuts them by roughly 27×. Each sigma level is a large quality leap, which is why the journey from 3σ (typical) to 6σ (best-in-class) is so consequential for cost of poor quality.
DPMO stands for Defects Per Million Opportunities. It is DPO (defects per opportunity = defects ÷ (units × opportunities per unit)) scaled to one million. By counting opportunities rather than just whole units, DPMO fairly compares quality across products of different complexity — a 50-part assembly has more chances to be defective than a single bolt, and DPMO accounts for that.
The 1.5σ shift is an empirical allowance for the fact that a process mean drifts over the long term by roughly 1.5 standard deviations. So a process that is 6σ capable in the short term performs like 4.5σ in the long term, which corresponds to 3.4 defects per million. The conventional Six Sigma DPMO table (including 6σ = 3.4 DPMO) already has this shift baked in.
The short-term Z is the actual number of standard deviations from the process mean to the spec limit, computed straight from the current defect rate with the inverse normal: Z = Φ⁻¹(1 − DPO). The (long-term) sigma level adds 1.5 to that to account for drift. So sigma level = short-term Z + 1.5. A 3.4-DPMO process has a short-term Z of about 4.5 but is called 6 sigma.
At the opportunity level, yield = (1 − DPO) × 100% = (1 − DPMO/1,000,000) × 100%. It is the percentage of opportunities that are defect-free. Note this differs from rolled throughput yield (RTY), which multiplies the first-pass yields of every step in a multi-stage process and is typically lower than any single-step yield.
Because defect reduction is highly non-linear. Going from 3σ (66,807 DPMO, ~93% yield) to 4σ (6,210 DPMO) cuts defects more than tenfold; 4σ to 5σ cuts them about 27-fold; 5σ to 6σ another ~68-fold. Small-looking gains in sigma level translate into massive reductions in scrap, rework, and cost of poor quality, which is why organizations invest heavily to climb even one level.
Try our Industrial & Systems Engineering Studio
More calculators, simulators, and guides for this discipline.