In mobile social networks, nodes are clustered by their interests and hobbies, and take part in some activities periodically. This paper puts forward an activity-based message opportunistic forwarding algorithm (AMOF) for the network characteristics. The main idea is that the biggest delivery probability node is selected and message is transferred to it, if the source node and destination node are both present in the same activities. While they are not in the same activities, the best link of the indirect delivery probability is found, and message will be transferred to it. Simulation results show that the proposed routing algorithm can not only improve the success of message delivery, but also reduce the network delay and overhead, compared with classical routing algorithms, such as Epidemic, PRoPHET, CMOT and CMTS.