Publicly available data of all hospital discharges has been available since Medicare changed to a case-based reimbursement system. A non-confidential version of this dataset contains a Zip Code identifier for each discharge, in addition to diagnoses, procedures, payer information, hospital charges and basic demographic data. The method for converting the raw data into a useful marketing database is described. An application of this database in conjunction with GIS is presented here. In this application, the market share of a community hospital is analyzed. A series of maps shows that geography plays an important role in hospital choice, and a linear regression model provides quantitative evidence of this pattern. Finally, bivariate maps are used for more complex analysis.