It is important that limited government resources be channeled to the poor, but it is not always easy to identify the poor. Which households should be given tranfers when reliable information on incomes is difficult to obtain? The authors of this paper present a simple method for targeting when income is not observable but other characteristics that are correlated with income can be observed. Using survey data taken from Cote d'Ivoire, they predict incomes based on observable characteristics and distribute transfers on the basis of those predictions. It appears that significant reductions in poverty can be achieved using this method.