Abstract—Routing is one of the difficult issues in mobile opportunistic networks, due to the lack of global knowledge and sporadic links. Related works use a greedy strategy to forward packets, i.e., they select relays with higher/bigger quality metrics. In this paper, we study the Opportunistic Routing Protocols (ORPs) from the perspective of social network analysis. Specifically, we discuss the impact of social relationship on the performance of ORPs. We first classify nodes into strangers and friends. We then explore the optimized number of strangers we can employ. Third, we propose STRON, a socially aware data forwarding scheme by taking both STRangers and their Optimized Number into account. We finally compare STRON with the state-of-the-art works through synthetical and trace-driven simulations, the numerical results demonstrate that STRON achieves a better performance, especially in terms of combined overhead/packet delivery ratio and the average number of hops per message.