User fees are used to recover costs and discourage unnecessary attendance at primary care clinics in many developing countries. In South Africa, user fees for children aged under 6 years and pregnant women were removed in 1994, and in 1997 all user fees at all primary health care clinics were abolished. The intention of these policy changes was to improve access to health services for previously disadvantaged communities. We investigated the impact of these changes on clinic attendance patterns in Hlabisa health district. Average quarterly new registrations and total attendances for preventive services (antenatal care, immunization, growth monitoring) and curative services (treatment of ailments) at a mobile primary health care unit were studied from 1992 to 1998. Regression analysis was undertaken to assess whether trends were statistically significant. There was a sustained increase in new registrations (P = 0.0001) and total attendances (P = 0.0001) for curative services, and a fall in new registrations (P = 0.01) and total attendances for immunization and growth monitoring (P = 0.0002) over the study period. The upturn in demand for curative services started at the time of the first policy change. The decreases in antenatal registrations (P = 0.07) and attendances (P = 0.09) were not statistically significant. The number of new registrations for immunization and growth monitoring increased following the first policy change but declined thereafter. We found no evidence that the second policy change influenced underlying trends. The removal of user fees improved access to curative services but this may have happened at the expense of some preventive services. Governments should remain vigilant about the effects of new health policies in order to ensure that objectives are being met.