Governments do not have enough resources to make anti-poverty programs available to everyone in many African countries. Targeting the poorest populations could increase the efficiency of available program resources. Anti-poverty programs could target either households or geographical areas. This study compares the potential poverty reduction impact of these two approaches in Uganda. The impact of various policy changes on the poverty rate in Uganda is simulated using the estimated parameters of an econometric model of household consumption. The policies examined are family-planning, increased primary school attendance, increased secondary school attendance, expansion of formal employment and micro-enterprise expansion. The results reveal gains in poverty rate reduction from program targeting as compared to randomly allocating finite program resources to a sub-sample of the population. Furthermore, geographical targeting is shown to be more effective than household targeting for most of the six policies examined in the study.