This report of a workshop on Supervisory Control was compiled under the Review of the National Research Council. Supervisory control is the human activity involved in initiating, monitoring, and adjusting processes in systems that are otherwise automatically controlled. The two-day workshop covered three major themes: (1) concepts and characteristics of supervisory control systems, (2) the choice of appropriate research vehicles, and (3) the interchange between researchers and designers. Summary conclusions are: (1) it is useful to characterize the emergency class of human-supervised, computer-controlled systems by strict as well as broader definitions. (2) no single or simple model of supervisory control is appropriate at this time. (3) experimenting with supervisory control systems is difficult for various reasons. (4) experienced subjects are essential for research. (5) supervisory control systems can never be completely closed, since the human supervisor must have the capability to set subgoals. (6) better guidance from researchers is needed for designers and operators, in the form of principles and checklists. Various articles cover the analysis of supervisory control systems and behavior, mental models, matching mental models of operators and designers and models of human performance. Failure modes are also discussed.