Automatic learning of network routing using random routes

A method and apparatus for automatically learning new routes within a distributed telecommunication switching system is disclosed. Randomly routed telecommunication calls are established from one of a plurality of originating switch nodes to one of a plurality of destination switch nodes over randomly selected telecommunication links not marked in an internal routing table of the originating switch node as being a route to the destination switch node. When each telecommunication call is established, a return message from the destination switch node provides routing information back to the originating switch node. Each originating switch node updates its routing table based on the results of the routing information returned when the random route is established. A random route is selected every predefined number of calls to each destination switch node.