Abstract: In order to solve the problem of a small number of sample data of products in reliability evaluation, this paper uses a Bayesian method to establish a reliability evaluation model for multi-source heterogeneous variable weight fusion. Firstly, this paper analyzes the tail gradient characteristics of the likelihood function of each information source and field data, and determines the weight of each information source through the absolute value error of the two. Then, considering the influence of the field test data on the weight of prior and multi-source fusion, this paper conducts reliability analysis of the product based on the Bayesian variable weight fusion method. Finally, the example shows that the variable weight fusion model is reasonable and effective, and the evaluation effect is better than the reliability evaluation model of Bayesian equal weight fusion.