Information overload problem in recent years makes the personalized recommendation technology arise, the collaborative filtering recommendation technology by establishing contacted between the user and the information has been widely used in every field of e-commerce. And in the field of the college entrance examination voluntary students also have the “information overload” problem, which means they cannot choose the suitable college from many colleges efficiently. Therefore, the idea of collaborative filtering is applied to this new field, take the students as users and colleges as the items in the recommendation system. By analysing students' voluntary reporting relevant data from the previous year, three processes of building user attributes matrix, finding the neighbor users and generating recommendation are described in detail. The recommendations results of the experiment show the effectiveness of recommendations systems, and it lays the foundation for further research work.