Opportunistic mobile social networks (OMSNs) are the network where mobile users utilize opportunistic contacts to transmit data by wireless peer-to-peer interaction. The growing share of using smart mobile devices offers the opportunity to build a ubiquitous infrastructure for data disseminations, so it is significant to study data transmissions in OMSNs. In order to improve the data dissemination performance of OMSNs, a data dissemination mechanism based on group structure (DDMGS) is proposed in this paper. Firstly, the relationship measurement model is designed based on the user’s behavior attributes that include the user’s movement trajectories, interests and communication behaviors. In addition, the group construction algorithm is designed for network topology composed of different behavior attribute relations. The topological structure based on location relationship has periodic stability. The topological structure based on interest relationship has long-term stability. The topological structure based on communication relationship has dynamicity. In order to improve the data dissemination performance and the overall network performance, a buffer management scheme is designed, and a cooperative game theory is introduced to strengthen the cooperation between nodes, to avoid the selfish behavior of the node. Simulation results show that, compared with the performance of direct delivery routing, prophetic routing, Simbet routing and Epidex routing, DDMGS has better performance in success rate of message transmission and the average hop count. It demonstrates that DDMGS is feasible and effective.