Social structure refers to relatively enduring patterns of behavior and relationship within social systems. In this paper, we propose a novel routing scheme, which takes advantage of the social structure of underlying 'owners' and their mobile devices in intermittently connected networks. We focus on routing by utilizing the social contact information. For example, device-owners come into contact with each other during attending regular meetings. We formulate the routing optimization problem, emphasizing the ever-changing network connectivity. Moreover, we propose an algorithm to infer the social structure by extracting contact information from online gathered traces. We further develop a routing protocol for intermittently connected networks based on that. The performance of our protocol is tested using real network traces. It is shown that the protocol is able to achieve a good delivery rate and latency, which is comparable with the optimal routing, and far outperforms the First Contact protocol. Our protocol is incrementally applicable to many current mobile ad hoc routing protocols.