SUMMARYProblems such as "DLL hell" that occur when multiple pieces of software are used on the same system cannot be detected by testing the functions of a single piece of software, but need to be detected by configuration testing. Configuration testing is performed as the final step in the development process, and although this means that there are many cases where there are not sufficient man-hours to fully conduct testing, existing methods may not be able to detect failures quickly. This paper therefore proposes a method for designing effective configuration tests called the resource path test method by focusing on failure mechanisms. Tools required in this method are presented, and the configuration test design process is explained. Examples of using the proposed method are also presented.