Abstract:  The idea of fault localization is general to locate the defective program statement via analyzing the program statement and its execution result. At present, there are many researches on software single fault location and few researches on multifault correlation. This paper constructs spectra based on program statement covering execution information, and proposes a software multifault location method based on clustering analysis. On the basis of Siemens testing cases set, the proposed method is compared with some existing localization methods based on program spectrum. In most cases, the software fault localization technology based on clustering analysis is verified to be more efficient than the methods in software multifault localization，and can reduce the tester cost.