The purpose of this paper is to evaluate academic database based on users' mental model theory. The evaluation indexes are developed by emotion aspects of users' mental model. The collected surveys (n=493), which are from academic database users, are analyzed by one-way analysis of variance method. Study results are as follows: Firstly, with the improvement of users' proficiency, the evaluation scores on macro orientation and retrieval method will also significantly improved; Secondly, on background system, information resource, information organization, retrieval interface aspects, and novice users' evaluation scores are significantly lower than intermediate-level users' and expert users' scores, but there is no significant difference between intermediate-level user and expert user; Thirdly, novice users' mental model is incomplete and inaccurate, and it will result in lower evaluation score. The poor design of academic database results in no significant difference between intermediate-level user and expert user.