Recommendation system focus on the individual recommendation and using relationships between users and users or between items and items by distance or similarity measures. In reality, there are many situations when recommending to individuals is not as important as recommending to groups. Researches on the relationship between a group and a group have not yet been interested in the recommendation. This paper mainly focus on applying the energy distance for group recommendation system. The proposed recommendation model is evaluated on the Jester5k and the MovieLens datasets. The experiment result shows the feasibility of applying the potential energy for the group recommendation problems.