Using both national-sample and program-level census survey data, we evaluate the distributional power of Mexico's Programa Nacional de Educacion, Salud y Alimentacion (PROGRESA) transfers using the so-called distributional characteristic. These transfers are targeted both geographically at marginal localities and at poor households within these localities. Transfers are also conditioned on household members attending school and health clinics. We show that the program has a relatively high distributional power compared to a range of alternatives considered. Although geographic targeting has a relatively large effect on the distributional power of the program, the demographic structure of transfers is more important than household targeting. However, the gains from household targeting increase as the program expands into less marginal localities. Within the structure of transfers, the education component is distributionally more powerful than the food component, reflecting the fact that the former is based on household demographics, while the latter is uniform across households. Restructuring education grants towards secondary schooling in order to generate higher education impacts does not appear to affect the distributional power of the program. In any case, any adverse impact could be offset by increasing the cap on transfers, which is regressive. Take-up of the program is high but relatively higher among the poorest households, thus increasing distributional power. However, this effect is mitigated by the fact that, conditional on program take-up, the poorest households take up a relatively lower proportion of potential transfers.