Routing protocol based on the social relationship in opportunistic networks can gain better performance in the opportunistic network with the social attribute. Several algorithms are proposed in literature for static social relationship in opportunistic networks, however these algorithms based on static social relationship cannot be used for the routing purpose in opportunistic networks as the social relationship in opportunistic networks are dynamic in nature. This paper investigates the problem of opportunistic network routing algorithms based on the dynamic social relationship. A new model of the dynamic social relationship in opportunistic networks is presented, and a new prediction method is introduced to predict the movement pattern and the encounter time. An optimized routing protocol in opportunistic network based on the dynamic social relationship is developed to rout messages. Our experiment results show that the proposed protocol is very efficient in the opportunistic networks with the dynamic social relationship, and the routing protocol in opportunistic is very suitable for the real application.