Simulated transfer schemes using household data for Venezuela, Mexico and Jamaica indicate that geographic targeting is a useful mechanism for transferring benefits to the poor. Poverty can be significantly reduced when compared with transfer schemes involving no targeting, such as general food subsidies. Of the various techniques used to identify priority regions for a targeted program, outcomes proved to be fairly similar. The level of geographic unit does, however, seem to have a notable impact on targeting outcomes. In the case of Mexico, the more narrowly defined the geographic region, the greater was the reduction in poverty. These findings have obvious implications for program design.