AbstractEarlier work has criticized the dominant tendencies in operational research contributions to health services planning as characterized by optimization, implausible demands for data, depoliticization, hierarchy and inflexibility. This paper describes an effort which avoids at least some of these pitfalls. The project was to construct a planning system for a regional health council in Ontario, Canada, which would take account of the possible alternative future states of the health-care system's environment and would aim to keep options for future development open. The planning system devised is described in the paper. It is based on robustness analysis, which evaluates alternative initial action sets in terms of the useful flexibility they preserve. Other features include the explicit incorporation of pressures for change generated outside the health-care system, and a satisficing approach to the identification of both initial action sets and alternative future configurations of the health-care system. It was found possible to borrow and radically‘re-use’techniques or formulations from the mainstream of O.R. contributions. Thus the‘reference projection’method was used to identify inadequacies in performance which future health-care system configurations must repair. And Delphi analysis, normally a method for generating consensus, was used in conjunction with cluster analysis of responses to generate meaningfully different alternative futures.