Delay tolerant networks (DTN) often use the copies of message to improve the success rate of message transmission. It can deduce the flooding of the messages in the network. So, the copies of message need to be limited. A social activity and copylimited based DTN routing algorithm (SACRA) is proposed. The algorithm defines the social activity based on the historical track and the movement characteristic of adjacent nodes. According to the social activity and the size of the unused buffer, the number of message copies and the exchange type of its carrying messages are distributed in a reasonable way. The message can be diffused across the network and carried to its destination node quickly. The simulation results show that SACRA can reduce the network load effectively while retain the message delivery ratio well. When the time is 1 000 s, the success rate is above 50%, and thus the network overhead is reduced and the success ratio of the message delivery is increased compared with other typical routing algorithms.