To date, the federal Head Start program has enrolled more than 31 million children from birth to age five, at an annual cost of nearly eight billion dollars. Given the magnitude of the Head Start program, its effects on children and their families are of obvious policy interest. In this paper, I seek to understand Head Start's effects on a specific group of enrollees: dual-generation Head Start participants. I ask whether the effect of Head Start differs for the three- and four-year-old children of former program participants, as compared to the children of non-participants, and find using experimental Head Start Impact Study data that the children of mothers who participated in Head Start receive larger programmatic effects on their cognitive test scores. I then try to determine the source of this difference by exploring several potential channels: sociodemographic characteristics, baseline cognitive differences, program "buy-in," and changes in maternal employment, maternal education, and parenting practices. I find that mothers' daily literacy practices improve by more if they participated in the program as a child than if they did not, suggesting that the improved parenting practices of former participant mothers drive the larger treatment effect on their children's cognitive skills. I also find evidence suggesting that this is a Head Start-specific phenomenon, not generalizable to other types of preschool. Overall, my findings indicate that first- and second-generation Head Start participation are complementary. (author abstract)