Summary: Aiming at the defect that the preference order value is difficult to be used to distinguish the preference intensity of the matching subject, a multi-stage two-sided matching method is proposed considering the staged characteristics of two-sided matching. Firstly, the intuitionistic fuzzy preference information of the matching subject is transformed into a basic trust allocation function. Then the matching weight optimization model is constructed according to the intuitionistic fuzzy similarity. The basic trust allocation function is modified and evidence is synthesized by using the matching weight. Furthermore, a matching satisfaction measure method is proposed based on the trust degree, and a two-sided matching optimization model is established. Finally, an example is given to verify the feasibility and validity of the proposed method.