A wide range of people are interested in how local factors influence patient choice among hospitals. Administrators need to know why patients are admitted to their hospitals so they can develop more sophisticated marketing of their services in an increasingly competitive environment. Planners concerned with issues of patient accessibility need know the geographic patterns of hospital use To meet these needs, it becomes necessary to develop methods to estimate the probability that patients will be admitted to a particular hospital using models that incorporate location and size of competing hospitals. In this paper, the focus of econometric investigation and prediction is the probability that a patient will select e particular hospital. Four different service areas were delineated and studied in West Virginia to test the Huff Consumer Spatial Behavior model for estimating demand at four hospitals that may be facing closure. It was found that through application of the Huff model that in a small system of hospitals and patients, each patient location (i.e., zip code) will send patients to nearly every hospital. The model predicted sufficient demand for two of the four hospitals studied. Conventional methodologies were then compared to the Huff model. The model did not test for financial feasibility of any of the facilities nor did the model adequately address the issue of how patients select a particular facility. / Master of Science