Modularity is one of the most prominent properties of real-world complex networks. Here, we address the issue of module identification in an important class of networks known as bipartite networks. Nodes in bipartite networks are divided into two non-overlapping sets, and the links must have one end node from each set. We suggest a novel approach especially suited for module detection in bipartite networks, and define a set of random networks that permit the evaluation of the accuracy of the new approach. Finally, we discuss how our approach can also be used to accurately identify modules in directed unipartite networks.