In view of the fact that it is difficult for the traditional fault tree model to express the multi-state and fuzziness of events,a reliability analysis method of complex system based on interval Bayesian network is proposed.Aiming at the problem of multi-state events in the fault tree model,the forward reasoning and reverse reasoning methods of Bayesian network are proposed to determine the probability indexes of the top event/bottom events in different states when the states of the bottom events/top event are determined.Aiming at the problem that it is difficult to determine the failure rates of bottom events,triangular fuzzy subset and interval triangular fuzzy subset are introduced to determine the reliability value of complex system under different input conditions.The accuracy and feasibility of the model is verified by the reliability analysis results,which has great practical value in engineering practice,and can be applied to the reliability analysis and fault diagnosis of complex system with small sample failure data.