Abstract: A method for clustering the Internet address space employs structural, topological, and temporal clustering techniques. Seedpoints are identified from among network destinations; the seedpoints are topologically clustered into groups; temporal measurements from one or more predetermined locations are made to a seedpoint in each group; and the seedpoints are clustered based on the measurements. The clusters are generalized based on information identifying the network addresses with seedpoints deemed to be close, such as address prefixes in a routing table. A representative is selected for each cluster, such as an intermediate node on a path shared by the seedpoints of the cluster. The technique can be employed by different types of applications, including route selection in an intelligent route controller.