This paper examines the tradeoff that consumers make between price and quality in the demand for health care. The analysis is based on data collected from both households and health care facilities in Cebu, Philippines. The availability of both types of data makes this one of only a handful of demand for health care studies that includes detailed information on both individual characteristics and facility attributes of all relevant alternatives. The developing country setting provides substantial variation in the type of facility chosen, ranging from home delivery aided only by friends and relatives at one extreme to modern private hospitals at the other end of the spectrum. The alternatives vary greatly in quality and price, making this an ideal context for examining the role of these variables in facility choice. The nested logit model specifications that are estimated contain price, travel time, and different combinations of quality measures, including the availability of medical supplies, practitioner training, service availability, facility size and crowdedness, and their interaction with individual characteristics. In addition, the sensitivity of the results to different choice-set definitions is analyzed. In particular, models that use conventional choice-set definitions that are based only on nominal status are compared with models that attempt to classify facilities into relatively homogeneous groups based on price and quality. The estimation results, which correct for the two-stage design of the household survey, indicate that facility crowding and practitioner training are significant determinants of consumer choice. The results also indicate that individual characteristics such as education of the woman interact in important ways with quality in influencing choice. For example, the availability of drugs is a significant determinant of facility choice for women with high levels of education, but not for others. In addition, the results support the hypothesis that price is a significant determinant for poor households, but not for other households. The model is used to conduct policy simulations designed to be informative to public officials interested in the effect of cost recovery schemes on utilization patterns. The simulations indicate that, when public facilities simultaneously increase user fees and the aspects of quality over which policy makers can exercise control in the short-run, the mean probability of using public facilities increases for both poor and non-poor households.