Free tool · updated 2026-10-23
Cpk calculator
Paste your measurements and spec limits to get Cp, Cpk, Pp and Ppk, with a 95% confidence interval for Cpk and the expected parts out of spec. It runs in your browser; your numbers are not uploaded.
How this calculator works
- Cp and Cpk use the within-process σ: the average moving range divided by 1.128 (MR̄ / d2), the standard estimate for one measurement at a time and Minitab's default for individual observations.
- Pp and Ppk use the overall sample standard deviation s of all your measurements.
- The 95% interval for Cpk uses Bissell's approximation (1990), the one in the NIST handbook. It needs about 25 or more measurements to be accurate.
- Expected out of spec assumes a normal distribution and uses the overall σ, so it estimates what the process actually produced, in parts per million (PPM).
The calculation runs the same code as the LimitLine app, published at /source. When this page is built, it is checked against the NIST/SEMATECH e-Handbook example: USL 20, LSL 8, mean 16 and σ 2 give Cp 1.00 and Cpk 0.67.
Before you trust the number
- Check stability first. Cpk assumes a process with one steady σ. If the control chart shows points outside the limits or clear patterns, the Cpk describes nothing in particular. This page doesn't check stability; the button above opens your numbers on a control chart.
- Use enough data. With 30 measurements a Cpk of 1.33 has a 95% interval of about ±0.36. See how many parts a Cpk needs.
- Look at the shape. The PPM figure relies on the normal curve. Strongly skewed data, such as flatness or runout bounded at zero, need a different method.
Questions
What is a good Cpk?
1.33 is a common minimum for an established process, and 1.67 is often asked for on new or safety-critical characteristics. Below 1.00, expect out-of-spec parts. Your customer or quality plan sets the target.
Why is my Cpk different from Excel or another tool?
Usually because of σ. Excel's STDEV gives the overall σ, which belongs to Ppk, not Cpk. Tools that estimate within σ from subgroups (R̄ / d2 or a pooled s) also give different results from a moving range. Check which σ each tool uses.
Can I use only one spec limit?
Yes. Leave the other box empty. Cpk is then the one side you have (Cpu or Cpl), and Cp isn't defined.
What's the difference between Cpk and Ppk?
Cpk shows what the process could do if it stayed as steady as it is from one part to the next. Ppk shows what it actually did. A Ppk well below Cpk means the process drifts or shifts over time. The Cp and Cpk guide walks through both with NIST data.
Sources
- NIST/SEMATECH e-Handbook of Statistical Methods, §6.1.6 What is Process Capability?
- A. F. Bissell (1990), “How Reliable Is Your Capability Index?”, Applied Statistics 39(3), 331–340
- NIST/SEMATECH e-Handbook, §6.3.2.2 Individuals Control Charts (the example data)