Abstract: Our purpose of the work in this paper is to obtain the relevant factor that influence the cultivation quality of graduate students from student pools,analyze the inherent relationships among these factors,and then evaluate the quality of student pools.Based on the probabilistic graphical model,we proposed the method for model construction.Then,based on the algorithm of probabilistic inferences,we gave the technique for quality evaluation of graduate student pools.By means of theoretical analysis,examples and experiments on real data sets,the feasibility of our methods have been verified.The proposal in this paper can provide an effective measure for evaluating the enrollment,cultivation and decision of graduate students.