One of the main reasons for allowing tax-exempt status to non-profit hospitals is the assumption that they will return a community dividend. Provision of health care for the poor and uninsured by these hospitals is a principal form of community dividend. Other forms of community dividend might include research, education, emergency room service, pediatric intensive care, etc. The purpose of this project is to analyze the relationship between the provision of community dividends, particularly charity care, by non-profit hospitals and the savings they obtain from tax exemptions. Due to their tax-exempt status, non-profit hospitals have access to less expensive capital. The subsidy they receive in the form of lower interest expense from tax-exempt bonds and savings from income tax on their net income might have some effect on the provision of indigent care. We would like to estimate the magnitude of the total savings from tax-exempt bonds and income tax and then study the relationship between the savings with the provision of indigent care. The study is important for two reasons. First, tax-exempt bonds and income tax savings generate a large subsidy from the community to the hospital. Second, its implication for access to health care for the poor and uninsured. The policy questions are whether those hospitals which have benefitted most from the subsidies are also the ones providing the maximum community services. If they are, it suggests that these subsidies are effective policy tools. If not, it may be appropriate for policy makers to consider making future subsidies explicitly conditional upon providing adequate community services. We shall perform our empirical work using data from the State of California, as we are already in possession of California's Annual Hospital Disclosure Reports for the time period 1982-88. The data set contains detailed information on hospitals' debt issues, expenses, revenues, operating and non-operating characteristics, payor mix, bad debt and charity care, service mix, staffing, and other pertinent data. We shall estimate our multivariate regression model for several measures of community services, namely, indigent care, Medicaid patient days, number of emergency room visits, presence or absence of a teaching program, and other relevant services.