Our objective is to test the hypothesis that aid can improve the welfare of the poor. Part of this effect is direct, if aid is targeted on the poor, and part is indirect, via the transmission channel of aid-financed public spending on social services – sanitation, education and health. This indirect part is represented in an index of pro-poor public expenditures (PPE). As comparative data on poverty levels are scarce, we use two indicators of the welfare of the poor, namely infant mortality and the Human Development Index (HDI). We use a residual generated regressor to obtain a coefficient on the aid variable that includes the indirect effects through public expenditure allocation induced by aid. Estimation is based on a pooled panel of 39 countries over the period 1980 to 1998. We obtain results in support of our hypothesis that ‘pro-poor’ public expenditure is associated with increased levels of welfare, and we find evidence that aid is associated with improved values of the welfare indicators because aid finances pro-poor spending. In this way, aid potentially benefits the poor.