QoS routing involves selection of paths for flows based on the knowledge at network nodes about the availability of resources along paths, and the QoS requirements of flows. Several QoS routing schemes have been proposed that differ in the way they gather information about the network state and select paths using this information. Most of these schemes can be categorized as best path routing where a source node selects the ‘‘best’’ path for each incoming flow based on its current view of the global network state. It has been shown that best path routing schemes require frequent exchange of network state, imposing both communication overhead on the network and processing overheads on the core routers. On the other hand, proportional routing schemes proportion incoming flows among a set of candidate paths. Two key questions that arise under proportional routing are how to select candidate paths and how to proportion flows among candidate paths. We propose a scheme that selects a few widest disjoint paths as candidates and equalizes the blocking probabilities of the candidate paths. We show that our proportional routing approach yields higher throughput with lower overhead than best path routing approach. Furthermore, we present a method for aggregating the state of an area and extend the proportional routing approach to provide hierarchical routing across multiple areas in a large network. 2003 Elsevier B.V. All rights reserved.