Sigma Level Calculator — Convert DPMO, Yield, Cpk & Sigma

A sigma level expresses process capability as the number of standard deviations that fit between the process mean and the nearest specification limit. The more standard deviations of headroom, the less likely a random unit falls out of spec: a 3σ process produces tens of thousands of defects per million opportunities, while a 6σ process produces just 3.4.

The scale comes with one famous wrinkle: the 1.5σ shift. Motorola observed that even a well-controlled process drifts over months, so the industry quotes two numbers. Short-term sigmais the headline value — the “6 sigma” in Six Sigma — which includes a +1.5σ credit for that drift. Long-term sigma is the raw observed z-value, exactly 1.5 lower. Every conversion table, certification exam, and vendor claim uses the short-term convention, and so does this converter.

Whichever input you provide, the converter first normalizes it to DPMO, then derives everything else from that single value — so the outputs always agree with each other and round-trip cleanly between modes.

σ (short-term) = table lookup from DPMO   (3.4 DPMO = 6σ, 66,807 = 3σ, …)
σ (long-term)  = σ (short-term) − 1.5

Yield = 1 − DPMO ÷ 1,000,000        DPMO = (1 − yield%/100) × 1,000,000

Cpk ≈ (σ short-term − 1.5) ÷ 3      (so Cpk 1.33 ≈ 4σ, Cpk 2.0 ≈ 6σ)

The sigma value is read from the industry-standard conversion table (with log-space interpolation between rows) rather than the continuous normal-distribution formula. The continuous formula runs about 0.2–0.3σ higher, which surprises anyone checking the result against the tables printed in Green Belt and Black Belt courseware — the table convention is what practitioners and examiners expect.

A call center resolves 92% of customer requests correctly the first time — a 92% first-time-right yield. What sigma level is that? First convert yield to DPMO: 8% of opportunities are defective, and 8% of one million is 80,000.

DPMO = (1 − 92/100) × 1,000,000 = 80,000
σ (short-term) ≈ 2.9σ   ·   σ (long-term) ≈ 1.4σ   ·   Cpk ≈ 0.47

80,000 DPMO sits between the 2.5σ (158,655) and 3σ (66,807) table rows, so the interpolated short-term sigma is about 2.9σ — a below-average process by industry norms. Reaching 3σ would mean cutting defects from 80,000 to 66,807 per million (about 93.3% first-time-right); reaching 4σ would mean a roughly 13-fold reduction to 6,210 DPMO.

These are the industry-standard values (including the 1.5σ shift) used in Green Belt and Black Belt training. The converter interpolates between table rows so its output matches what you would look up by hand.

Most businesses run at 3σ to 4σ; hearing that your process is “only” 2.9σ is normal, not alarming. What matters is the trajectory: each sigma level represents roughly an order-of-magnitude fewer defects, so moving from 3σ to 4σ is a much bigger (and more valuable) jump than the arithmetic suggests. Use the converted DPMO to size the gap — “we need to eliminate 9 of every 10 defects” is a clearer improvement target than “gain one sigma”.

Two cautions. First, sigma computed from a small sample is fragile: zero defects in 50 units does not make a six-sigma process — it makes an insufficient sample. Second, always say which convention you are quoting. A “4σ process” short-term is a 2.5σ process long-term, and mixing the two in one report is the fastest way to a confused steering committee. This page always labels both.