A teleconferencing system in which a teleconference controller advantageously uses information relating to the network topology layout of remote nodes which are to participate in a teleconference, and then advantageously directs the transmission of video and/or audio streams from a source node to an identified subset of the remote nodes based on the capabilities of the nodes and the network links. These nodes which initially receive the video and/or audio streams then advantageously forward the streams in turn to other nodes in their vicinity using peer-to-peer network connectivity techniques. The identification of the subset of nodes which serve as peer-to-peer forwarding nodes may be based on node activity, node service capacity, the impact on network links between the node and the requester, the distance and/or latency between the node and the requester, and the ability of the node to meet requests including video transcoding requests.