The reasons for the persistence of the academic department in research universities and its efficacy in achieving the typical goals of the institution are examined. Six reasons that the academic department in its present form persists as an organizational entity are: inertia, the high status of research, funding sources and access to them, insulation from observation and accountability, alleged interdependence among missions, and the pedagogical necessity for knowledge-based departments. Various theories of departmentalization, including current notions of the determinants of alternative organizational structures as posed by "contingency theorists," are examined. The nature of the environments external to the institution and the varying demands of clients in those environments for different structures are explored. The impact of the "technology" of the many university tasks on the structure of the institution is discussed. A matrix form of organization in which sets of tasks or roles which are related to one another by necessary timing and sequencing are grouped in self-contained units is proposed. In such units, all the major resources needed to provide the service or produce the output are contained within the unit. It is suggested that the amount of inter-unit dependency should be as low as possible in order to minimize the costs of administrative coordination, and the amount of intra-unit loyalty and identification with unit goals and norms should be increased. (SC)