Game-based learning is an effective learning method, whose performance depends on the quality of the educational game. Due to versatile game environments with complex backgrounds, evaluations are not easy to implement. Consequently, it is difficult for educators to determine to what degree a game may be qualified. This study proposes a novel, multiple-criteria decision making (MCDM) model based on Flow theory to help educators evaluate educational games. This study designs a Flow dimension questionnaire in which some domain experts are asked to re-examine the attributes of the Flow dimension by using the Delphi method. The extracted attributes of Flow dimension are then employed to build the proposed model for evaluating a given educational game. For practical implementation, this study developed an Evaluating Mobile-Learning Game System (EMGS), based on the proposed model. In the case study, several experts were requested to use the developed system for evaluating three different types of educational games. After the evaluation process, the experts were invited to fill out a questionnaire of system satisfaction. The results show that the developed system is faster and more convenient than the traditional evaluation method, and the developed system could effectively recommend the best qualified educational game for educators in different situations.