Process capability is the measurement of a property and comparison of that property to the specification limits. Such measurement and comparison is calculated using capability indices. The result of the comparison is usually represented by a histogram.

The output of every process has at least one measurable quality which can be statistically analyzed. Standard deviation and process means can be used to describe the process if it depicts a normal distribution. A process control chart is used to check if the process is statistically in control. Therefore the first step in process capability analysis is to collect as much as data as possible. This output must meet a specification, requirement or a product tolerance.

Capability study produces four indices: the Cp index, Pp index, Cpk index, and Ppk index. Cp index is the process capability index; Pp index is process performance index; Cpk index is a kind of process capability index that is used in statistical process control and Ppk is a process performance index used in statistical process control. Cp is useful when the data is affected by special causes and can be used to predict the process capability if no such uncertainties existed. Ppk, on the other hand, provides the most realistic estimate of the capability of the process. A Ppk up to 1.0 would mean that the process needs to be improved and a Ppk up to 1.5 would mean that the process is excellent.

Some of the reasons why process capability should be measured are to prioritize different processes, to investigate the causes of process failure, establishing specification limits, and following-up on the improvements that were made. A process capability study can be performed in the following manner:

1. Selection of a candidate
2. Definition of the process
3. Preparation of resources
4. Evaluation of the system of measurement
5. Preparation of control plan
6. Selection of method of analysis
7. Gathering and analysis of data
8. Identification and removal of special causes
9. Estimation of process capability
10. Preparation of a follow-up plan

Each of the indices is calculated based on the assumption that the process distribution takes a bell shaped curve or is normal. The data collected for the study purpose must be inclusive of all natural variations. If the data does not show normal distribution it can be converted to the normal form using Box-Cox Transformation. A new set of indices can also be developed for non-parametric data.