Critics of user charges for government primary health care have focussed on the deterrent effect these charges might have on the poor, but there is little convincing empirical evidence that supports or contradicts these claims. The main research aims were to assess the equity effects of user charges for curative PHC services on households in 2 rural districts of Sierra Leone, a country that has suffered severe economic hardship in the last decade. Secondary objectives were to assess the feasibility of using objective means-testing to identify patients for exemption, and to recoimiend a simple methodology for acquiring the same information for local, operational purposes. A survey of 1156 households was carried out in the dry post-harvest season, and covered a range of household economic factors in addition to the actions taken in response to all reported illness episodes. A followup survey was made the following rainy season to assess seasonal effects. Supplementary information was obtained through focus groups and case studies. The data were analyzed within the framework of a conceptual model which assumed that preferences, access, and ability to pay were the main factors (or groups of factors) that determined which of several medical and non-medical treatment options would be used. Multiple regression models were used to assess the effects of each group of factors. The main findings were that, while wealthier households used cheap market drugs and expensive medical treatment options more than the poor, there was little difference in use of medium-priced PHC treatment. Household wealth and income factors correlated weakly with amounts actually paid for treatment. The immediate availability of money in the household appeared to be the economic factor most affecting utilization, with wealthier households nearly as likely = to have the amounts needed for PHC treatment on hand as poorer ones. Distance was a much more important determinant of choice of treatment than was income or assets, as were certain preference factors. However, the poor spent a much higher proportion of household income on treatment than the wealthy, so a way of limiting total expenditures for the poor would be more important than limiting their deterrence. Several readily-ascertained household factors correlated well with household income, but means-testing was concluded to be an inefficient way to accomplish the objective of selectively limiting expenditures unless incorporated into a prepayment scheme.