Abstract: The MDS-MAP(P) localization algorithm has shortage of calculation errors of the shortest path distance, combined error, and the higher complexity of the algorithm. In order to solve these problems, A density-based clustering localization algorithm called MDS-MAP(DB) is proposed. The algorithm selects a node which has most neighbor nodes as the start node in the clustering mechanism. The clustering which is built by 1 hop neighbor nodes uses the triangle inequality estimating missing distances. The clustering which is built by 2 hop nodes uses the shortest path distance estimating missing distances. In all clustering, only the cluster head executes the algorithm of estimating distance. In this way, the error of estimating distances and the computational complexity of this algorithm have been reduced. The results of simulation show that the proposed algorithm can reduces the average estimation errors.