Opportunistic routing is a new design trend of wireless routing, which can dramatically improve the end-toend throughput over traditional routing by judiciously utilizing the broadcasting nature of wireless network. However, the global scheduling scheme it adopts restricts its application in large-scale wireless network, due to the big waste of the end-to-end transmission latency and the computation cost. So, how to schedule the transmissions of different forwarders is essentially an important issue of the opportunistic routing scheme design. In this paper, we propose a graph partition based local scheduling (GPLS) scheme for opportunistic routing. It partitions the wireless topology into several subgraphs, and realizes the local forwarding in each subgraph. Simulation results show that our local forwarding scheme can significantly improve the network performances in large-scale network over existing opportunistic routing schemes, in terms of the end-to-end delay and the life time of the wireless node.