A technique for discovering the topology of a computer network and which relies on initially resolving the positions of end stations on the network and subsequently resolving the topology of remaining devices. This technique specifically involves querying a port of a managed device for a first address of any packet which it received; querying the same port for a second address of any other packet which the device received; if there is no such second packet, then returning data to a manager and determining if there is another port on the device; if there is such a second address, determining if there is another port on the device; repeating the preceding steps for the other port of the device; and if no such other port exists, repeating all the preceding steps for another managed device on the network.