Abstract:A fault tree of CNC systems was developed by functional decomposition, which took a representative instance for thread machining on digital controlled lathe. The fault tree regulation was demonstrated by case-based reasoning. A method to infer failure cause of CNC lathe on the basis of maximum degree of membership was established. By using in combination the failure data collected from maintenance of the factories manufacturing machine tools, the malfunction intensity matrix was set up based on control theory of fuzzy failure tree, which was able to use to decide degree of membership about a failure mechanism. This method can improve the working efficiency and veracity of system fault diagnosis of CNC machine tool, which also can offer a new inference method to design expert system for fault diagnosis.