There is provided a process analysis method capable of clustering of executed processes and extraction of similar processes.A process analysis method of the present invention converts the similarity between processes into a numeric value as a metric by including a process matrix transform 12 and a metric calculation 14 for process records 11 made up of a series of records of rendered service of at least three elements of the time of service, the type of service, and the quantity of service. This enables the calculation of the similarity between processes as a metric from information of only a series of records of rendered services. Accordingly, it becomes possible to cluster the executed processes and to extract similar processes based on the metric herein calculated.