This paper explores the issues and pitfalls encountered when attempting to test empirically the hypothesis that physician, hospital, or any other input supply level induces increasing demand for health services in the strict sense of demand shift and, through that, increased demand for the input in question. Evidence is presented which suggests that an empirical test of the supplier induced demand (SID) hypothesis of the type traditionally performed may not in fact be feasible with cross-sectional aggregate data such as is usually used.