It is conducive to ensure the safety and reliability of the power system by means of timely and accurate evaluation of the reliability of large generators.An evaluation method of fault probability for large generators based on multi-characteristic parameters is proposed in this paper,which can be employed to solve the difficult problem of evaluating the occurrence probability of fault modes of large generators in the process of reliability evaluation.Firstly,the set of fault characteristic parameters for a large generator is established on the basis of a large number of operation experiences and preventive test code for the large generator.Secondly,according to the features of the fault characteristic parameters,two estimation methods of the fault mode occurrence probability are proposed,which are based on the fuzzy membership function and the relative deterioration degree,respectively.Finally,on the basis of the improved analytic hierarchy process（IAHP） method,the composite estimation method of the fault probability for a large generator is presented,which can reasonably and adequately use the information of various characteristic parameters.An assessment case is provided to illustrate the specific process of this evaluation method and verify its effectiveness.