To enhance the interpretation and inference ability of the uncertainty in the process of machining defect diagnosis, Bayesian networks is introduced into the diagnosis process. The machining defect diagnosis method and inference procedures based on Bayesian Networks are proposed. On the one hand, it takes actual machining conditions and defect phenomena as complex evidences to improve diagnosis accuracy; on the other hand, it is possible to get the probability of each factor and to find out the maximum probability path using Bayesian Networks inference. Then a diagnosis method towards uncertain information is provided. At last, a case study for the machining defect diagnosis of the rotor flange connected hole is reported to illustrate the proposed method.