The evaluation of applicants' learning potential is important to college admission process. This paper develops a multi-criteria decision making method for a comprehensive evaluation of high school graduates' learning potential in college, in which both entrance examination marks and expert remarks are considered in the indicator system. Experts' opinions towards indicator importance are expressed by hesitant fuzzy numbers. By using hesitant fuzzy linguistic judgments, the flexibility of expressions is increased. A minimized divergence and hesitant degree model is established to calculate the experts' weights. Then to determine indicators' weights, a weighted average operator is applied. To aggregate the final evaluation results, a TOPSIS method is adopted. The proposed methodology is applied to evaluate students' learning potential in one of the top universities in China. Ten years' real enrollment data is collected in Business and Economic School. Based on the evaluation results, some suggestions are given for the admission process.