The National Biomedical Communications Network has evolved both from a set of conceptual recommendations over the last twelve years and an accumulation of needs manifesting themselves in the requests of members of the medical community. With a short history of three years this Network and its developing structure have exhibited most of the stresses of technology interfacing with customer groups and of a structure attempting to build itself upon many existing fragmentary unconnected segments of a potentially viable resource-sharing capability. In addition to addressing these topics, the paper treats a design appropriate to any network devoted to information transfer in a special interest user community. It discusses fundamentals of network design, highlighting that network structure most appropriate to a national information network. Examples are given of cost analyses of information services and certain conjectures are offered concerning the roles of national networks. (Other papers from this conference are available as LI 003360 - 003387 and LI 003389 - 003390) (Author)