With respect to the determination problems of threshold value on multi-attribute grey incidence clustering, based on the method of decision-theoretic rough set, by introducing two threshold parameters, the concepts of possible relationships and the classes set among objects are defined, which are exploited to substitute the one or the other relationship of grey incidence clustering method, and then a multi-attribute grey incidence clustering model based on decision-theoretic rough set is constrated. Further, Bayesian reasoning is utilized to study the computing mechanism of threshold value based on multi-attribute grey incidence clustering. Finally, an example is used to illustrate its feasibility and effectiveness. The results show that the proposed model is the expansion and generalization of the classic grey incidence clustering method, and can objectively and scientifically determine the threshold value of multi-attribute grey incidence clustering method.