This paper examines the effect of Brazil's Family Health Program (Programa Saude da Familia, FHP) on cognitive skills of fifth-grade students. We use biennial data from national exams between 2007 and 2015, and variation in the FHP implementation date across municipalities, birth cohort, and test year to identify the effect of the program on language and mathematics test scores. We find that, in northern municipalities, students exposed to FHP at or prior to birth have 0.88 points higher language and 1.30 points higher mathematics test scores compared to those exposed to FHP in childhood. The estimated effects are intent-to-treat effects and correspond to increases of 0.021sd and 0.030sd in language and mathematics test scores. We use an event-study analysis demonstrating that the largest effects of FHP on cognitive skills are for those students exposed at or prior to birth, with trivial effects if exposed after birth. We do not find evidence for changes in parental investment behavior or child school attendance, which suggests that the effects are likely due to the direct impact of the program on child cognitive development.