Location-based recommender systems (LBRS) suggest friends, events, and places considering information about geographical locations. These recommendations can be made to individuals but also to groups of users, which implies satisfying the group as a whole. In this work, we analyze different alternatives for POI group recommendations based on a multi-agent system consisting of negotiating agents that represent a group of users. The results obtained thus far indicate that our multi-agent approach outperforms traditional aggregation approaches, and that the usage of LBSN information helps to improve both the quality of the recommendations and the eficiency of the recommendation process.