Governments in developed and developing countries alike typically subsidize social services such as health and education, partly in recognition of the broad social benefits that are generated, and partly to serve broad equity goals. Governments often provide such services themselves, or at least play a major role in provision, and subsidy is often effected by setting prices at government facilities at levels that are considered affordable by the poor. The extent to which redistribution is achieved, however, depends on how well proper targeting is accomplished. The article uses a number of public hospitals in Thailand to explore the pricing of hospital services and its effect on provider and user behavior. A survey of hospital users helps to identify the beneficiaries of the current system. The degree of cost recovery for each hospital department public facilities is calculated, and user responses to price and income changes are estimated. Costs, fees and actual payments are compared to provide an indication of the extent of subsidy and the degree of cost recovery from different groups of patients, and it is found that although public hospitals appear to be the only place affordable by the lowest income groups, they also serve middle to high income users as well. It is argued that targeting can be improved and that hospitals charge only about 52-58 per cent of cost for out-patients, and 62-66 per cent of cost for inpatients. In addition, among those who finance their own health care expenses, actual payment varies from 20 to 90 per cent of charges. As for the extent of subsidy, an out-patient in the North receives an 88 per cent subsidy as in-patients receive an even higher subsidy of 94 per cent. There is also some evidence that the extent of subsidy is high for the higher income groups. While welfare programs appreciably benefit the lowest income groups, this is not exclusively so, and many beneficiaries fall into the medium and even high income categories. Government officials, most of whom are classified in the middle to high income groups, turn out to be most heavily subsidized through the civil service reimbursement of medical expenses. On the whole the estimated demand elasticities show that the demand for in-patient and outpatient care is in general not sensitive to changes in price, money income, or total household income. Further investigation into the possibility of greater selective reliance on user fees, with built-in safeguards and better targeting of public services to protect the disadvantaged groups, is recommended. Specifically,a closer alignment of user fees with the costs for services utilized by higher income groups (such as private rooms), with provision for continued subsidization of low income groups is recommended. Targeting would thus be improved.