Abstract: Owing to the long-life and small sample characteristics of the momentum wheel,the large sample life experiments to assess its reliability is carried out.But there is a great deal of prior information obtained from specialists and ground testing,so a new momentum wheel reliability modeling and assessment method using Bayesian network is presented.First,the momentum wheel's reliability model is built using various kind of prior information by Bayesian network learning algorithm.Then,the way of assessing the reliability of momentum wheels is provided and the faults is analyzed using the inference algorithm of Bayesian network.Finally,some proper examples are provided to prove the method's validity.