Abstract: In this paper, a new kind of clustering algorithm that is called DCUR (density-based clustering using k references) is presented. Innovative point is that the new algorithm to k references response data Distribution, and then analyzes the data based on the k references. DCUR keeps the ability of density based clustering method's good features, and it can reach high efficiency and the execution frequency of region query can be decreased, and consequently the I/O cost is reduced, so the new algorithm can reach high efficiency and reduce the complexity of the algorithm. Both theory analysis and experimental results confirm that DCUR is effective and efficient in clustering large-scale database, and its executing efficiency is much higher than traditional DBSCAN algorithm based on R*-tree.