This paper presents an online fault diagnosing method for permanent magnetic DC motor. Fault model are defined by analyzing dynamic mathematical model of the permanent magnetic DC motor in loading status. According to the fault features defined in this paper, the author investigate failure experiments on several motors with the most common faults in application and analyzed the fault mechanism theoretically. The consistency between experimental results and the analysis in theory shows that the method presented in this paper is feasible for detection online faults of permanent-magnetic DC motor.