To improve the overall comfort of manned closed cabin, enhance the physiological and psychological comfort and work efficiency of operators, an in-depth study on evaluating the comfort level of manned closed cabin is conducted. Based on the comprehensive comfort evaluation of the cabin, the basic events affecting the comprehensive comfort of the cabin are identified with their values ranked in the order of importance so as guide the design of the manned closed cabin in a highly specific way. A fault tree analysis (FTA) model was advanced for the comfort of the manned closed cabin. The influencing factors of the comfort of the manned closed cabin were divided into three intermediate grades: physiological environment, physical factors and subjective feelings, with 17 basic events were extracted. The fuzzy Bayesian networks (FBN) was used for forward diagnosis to evaluate the overall comfort of manned closed cabin, and reverse causal inference was performed to identify the main causes for the discomfort of the cabin. In the case study, the overall comfort level of the manned closed laboratory of Northwestern Polytechnical University was evaluated, the validity and reliability of the FTA-FBN comfort evaluation approach for manned closed cabin was verified by the subjective evaluation of 16 participants. The influence probability of each basic event on the overall comfort of the cabin was calculated by backward reasoning and sorted. The improvement direction of the future closed cabin design was suggested, providing a fresh insight into the comfort evaluation of manned closed cabin.