This paper shows that a UD (Usage Distribution) coverage criterion based on an operational profile measures the progress of software testing from users’ viewpoint effectively. The UD coverage criterion focuses on not only the structure of software but the usage characteristics of users. Its measuring object is the software usage weighted with the probability of the operational profile, and intuitively it shows what percentage of expected use in operational environments has been tested. It solves the problem that usual coverage criteria focussing only on the structure of software are not substantiated by software reliability and do not identify the degree of an importance of each measuring object. This paper describes the definitions and examples of UD coverage.