In the reliability evaluation of cycloid gear grinding machine,aiming at the problem of less field test data,a Bayesian reliability evaluation method considering the multi-source of prior data was proposed.A mixed prior distribution was established by using the credibility between different samples,and then the multi-source prior information was fused by using the Bayesian method.The fusion information was used as the integrated prior information by combining a likelihood function of the simulation data,the posterior distribution of the parameters was obtained.The simulation example shows that the MTBF point estimation value obtained by this method is 997.37,and the MTBF error is 3.2 %,which is smaller than that obtained by traditional Bayesian method.The fitting reliability curves are closer to the actual conditions,the proposed method is more accurate in the reliability evaluation of the gear grinding machine.