Abstract: The probabilistic safety analysis (PSA) results of the nuclear power plant shows that common cause failures (CCF) plays an important role in the system reliability. In China, generic data are often used for common cause failure data in PSA. It is difficult to reflect the operation characteristics of the nuclear power units. The alpha factor model is the most commonly used model in PSA to model the common cause failure due to its simplified parameter estimation form and accurate calculation results. However it is difficult to obtain reasonable statistical values with the classical estimation algorithm due to the rarity of common cause failure events. So the paper introduces the Bayesian estimation algorithm. The method can combine the prior information and sample information to obtain a better estimation without a large sample. The problem of lack of common cause failure events in nuclear power plants and unreasonable calculation results by classical estimation method are solved effectively. The method is suitable for the parameter estimation of common cause failure model in nuclear power plants.