The health sector consumes ever growing shares of the GDP in most Western countries. This is a reason for political concern and leads to many discussions on health system reforms. A keystone for the evaluation of health systems and suggested or realized reforms is an understanding of the determinants of health care utilization. Estimation of the demand for health care can help in forming predictions for future trends in health care utilization which can be caused by changes in the socio-demographic structure of a population. It is also of help in developing health care reform proposals and enables the evaluation of realized reforms and policy measures. The importance of the topic, the growing availability of health-related micro-data and my interest in applied microeconometrics led to this thesis. It contains three chapters which are linked together on three dimensions. First, all chapters are concerned with estimating determinants of health care utilization. The three chapters give comprehensive information on determinants of ambulatory health care utilization in Switzerland. Second, econometric methods are applied which have been developed to estimate count data models with endogenous regressors. Health service utilization – measured by the number of physician visits or hospital spells – is a prime example for count data. The estimation of health care utilization should take into account that many of the explanatory variables cannot be considered to be exogenous. Health status and health care utilization are interdependent and the choice of a specific health insurance contract is made based on expected health care utilization. Third, all chapters have an evaluation component. Chapter one estimates health care utilization of elderly Swiss in the context of a disability prevention study. It is one of few studies with availability of an extensive longitudinal database on both health, socio-economic and health care use variables. New insight are gained into the mechanism of preventive programs for older people. As hypothesised, the data show that the preventive intervention results in an initial increase of health service use, followed by a subsequent decrease if the intervention is effective. Chapters two and three estimate incentive-induced changes in physician service utilization by choosing a higher deductible than is minimally required. The results indicate that the effect of choosing a higher deductible is overestimated when endogeneity is not controlled for. Most of the observed lower number of physician visits can be attributed to self-selection of low risk individuals into insurance contracts with higher deductibles.