A novel breadth-first based structural clustering method for graphs is proposed. Clustering is an important task for analyzing complex networks such as biological networks, World Wide Web (WWW) and social networks. The clusters are of various shapes such as cliques and stars, for example, in Protein-Protein Interactive (PPI) networks. Traditional algorithms may detect clique-shaped clusters, but they fail to identify star-shaped clusters that are common in scale free networks, including PPI networks. The authors propose a novel clustering algorithm to solve the problem.