Abstract: The risks of subway tunnel structure in operation are complex and influential. How to efficiently evaluate the risk level of subway tunnel structure in operation and how to quickly identify the key risk factors to ensure the safety of tunnel structure need urgently to be solved. Thus, a risk assessment method based on fuzzy multi-state Bayesian networks is proposed. Firstly, the establishment process of fuzzy multi-state Bayesian network model is expounded. Then, the Bayesian theorem is introduced to predict the probability and analyze the influence factors of operational subway tunnel structure to evaluate the system risks. Finally, a case study is carried out to demonstrate the applicability of the proposed method which provides a reference for risk prediction and prevention of the tunnel structure of subway.