This article is part of the CNC Tolerances & Quality Control Guide: GD&T, CMM, Cpk, PPAP on CNX Precision.
Tolerance alone never tells the whole story of a machined part program. Two shops can hit the same blueprint dimension and still deliver very different consistency. The Cpk capability index quantifies that consistency. It compares the spread of measured values to the tolerance window and shows how close the process runs to its limits. For buyers of CNC machined parts, Cpk turns a vague quality question into a number that can be tracked, compared, and enforced.
What Cp and Cpk Measure
Cp and Cpk are process capability indices. They describe how well a stable process fits inside a tolerance zone. Cp looks only at spread. It divides the tolerance width by the natural variation of the process, expressed as six standard deviations. A high Cp means the process variation is narrow relative to the tolerance. Cp says nothing about centering. A process can carry an excellent Cp while drifting toward one tolerance limit and still produce scrap.
Cpk adds the centering check. It measures the distance from the process average to the nearest tolerance limit, expressed in units of three standard deviations. When a process sits exactly at the tolerance midpoint, Cp and Cpk are equal. As the average drifts toward either limit, Cpk falls below Cp. That gap is the warning signal. The parts may still pass inspection, but the margin is shrinking. For machined dimensions such as bore diameters, shaft journals, and true position values, Cpk captures precision and aim in a single figure.
How the Index Is Calculated
The math is straightforward. Cp equals the upper specification limit minus the lower specification limit, divided by six times the standard deviation of the measured data. Cpk equals the smaller of two values: the upper limit minus the process mean, divided by three standard deviations, or the mean minus the lower limit, divided by three standard deviations. The calculation assumes the data follow a normal distribution and come from a process in statistical control. Measurements must be taken with calibrated instruments, and gage error should be small relative to the tolerance.
A capability study defines the inputs. Shops typically machine a run of consecutive parts, often thirty or more, under normal conditions with no sorting or adjustment between samples. Each part is measured with the same instrument and the same method. The resulting data set produces the mean and standard deviation that feed the formulas. Some studies run over longer periods to capture shift-to-shift and lot-to-lot variation. Longer studies tend to report lower capability than short snapshots because they include more sources of variation.
Interpreting the Cpk capability index: 1.33 and 1.67
Industry practice has settled on common benchmarks. A Cpk of 1.33 is the widely accepted minimum for production characteristics in many automotive, aerospace, and industrial programs. It means the process spread fits comfortably inside the tolerance with margin on both sides. A Cpk of 1.67 is typically required for critical or safety-related characteristics where the cost of a defect is high. Some programs demand 2.0 for the most sensitive features. Requirements always come from the drawing, the contract, or the customer quality manual, so buyers should state them before quoting. Because every program sets its own bar, the Cpk capability index only has meaning against an agreed requirement.
What do these numbers mean at the spindle? A Cpk of 1.33 on a shaft diameter suggests the process is well centered and unlikely to produce scrap under normal variation. A Cpk below 1.0 means the process spread approaches or exceeds the tolerance band, so scrap becomes a matter of time rather than luck. When the Cpk capability index reads low, the shop must either reduce variation, recenter the process, or negotiate the tolerance with engineering. Ignoring a low value simply moves the problem downstream to incoming inspection or, worse, to the field.
How Shops Demonstrate Capability and Why Buyers Should Ask
Serious shops prove capability instead of asserting it. The most common evidence is a capability study attached to the first article inspection report or PPAP submission. The study lists the characteristic, sample size, measurement equipment, calculated Cp and Cpk values, and a histogram of the data. Statistical process control extends the proof into production. Operators or automated probes measure parts at defined intervals and plot results on control charts. When a trend crosses a control limit, the process stops for adjustment before bad parts accumulate.
Modern CNC lathes and machining centers support this workflow directly. In-process probing measures features between cuts and applies automatic tool offsets, while tool wear compensation keeps dimensions stable across long runs. Calibration records for micrometers, bore gages, and CMMs back up the data. Each capability report states the resulting Cpk capability index so reviewers can compare runs over time. Buyers benefit from asking about capability early. A supplier that tracks Cpk exposes process health before defects do, and incoming inspection loads drop because the process demonstrably holds the print. Quoting discussions become concrete: if a tolerance cannot be held at the requested capability level, the shop can propose a design adjustment rather than absorb silent scrap costs. For high-volume programs, requiring a minimum Cpk at approval and monitoring it in production is one of the cheapest quality insurance policies available.
Where Cpk Falls Short
The Cpk capability index is powerful, but it has limits. The formulas assume a normal distribution. Bimodal data from two cavities, two spindles, or mixed lots can produce a misleading single value. Cpk also assumes stability. A process drifting with tool wear can show an acceptable Cpk in a short study while producing scrap a week later. Measurement error matters too: if gage repeatability consumes a large share of the tolerance, the index reflects the instrument as much as the machine.
Cpk is also a statistical statement, not a substitute for inspection. A high value reduces risk but never eliminates it, and some contract features still require 100 percent inspection regardless of capability. The number hides root causes. It flags that a problem exists but not whether vibration, thermal growth, or material hardness drives it. Finally, Cpk says nothing about attributes that cannot be measured as variables, such as visual criteria. Use it as one input to supplier evaluation, not the only one.
Frequently Asked Questions
What is a good Cpk value for machined parts?
Most programs accept a minimum of 1.33 for standard characteristics. Critical and safety-related features commonly require 1.67, and some OEMs specify 2.0. Always confirm the required value in the drawing or quality agreement before production starts. A low Cpk capability index on a critical feature should pause production until the process is corrected. The right target depends on part function, not on a universal rule.
What is the difference between Cp and Cpk?
Cp measures only process spread relative to the tolerance. Cpk also accounts for centering. The two values are equal when the process average sits at the tolerance midpoint. A large gap between them tells you the process is off center and needs adjustment before capability claims are made.
Can Cpk be calculated from first article inspection data?
Technically yes, but a first article usually measures too few parts for a reliable study. Capability studies need consecutive parts produced under stable conditions, typically thirty or more. If a supplier quotes a Cpk from five samples, treat it with caution and request a proper capability study.
For related information, see our guide to cnc machining service and 5-axis cnc machining and cnc machining tolerances, and aluminum cnc machining.