Most discussion of after-school programs in shantytowns has centered on estimating mean impacts of programs, and results are not conclusive. Previous literature provides some explanations for these mixed results but this paper provides a new channel: the effectiveness of an after-school program on students depends on their parents’ type. One can argue that those parents who live in a shantytown may be there due to their bad type or because of bad luck (good type parents who are in a shantytown because they had bad luck in their lives but if they received an opportunity –such as an after-school program for their children- they would exploit it). The complementarities between after-school and parents’ good type are not obvious. Is a good policy to suggest responsible and committed parents to leave their children many hours a day in an after-school program? Would be better for those children to remain at home in contact with their committed parents? Should policy be directed to the children of bad type parents? By using random assignment to evaluate an after-school program in a developing country, we find that it is effective in raising children’s school achievement for those whose parents are of good type. Thus, this paper provides evidence that the knowledge of the distribution of impacts is crucial to guide public policy and it is not enough just to change the environments in which youth spend their afterschool hours, increasing time in safe, supervised settings.