We study how fathers' time impacts children's human capital using the introduction of earmarked paternity leave in Sweden. We use administrative data on parents' leave uptake and children's educational outcomes in a difference-in-discontinuities design, exploiting the plausibly random timing of childbirth. We show that the reform decreased average school-leaving grade point averages of sons of non-college fathers by 0.07 standard deviations and increased intergenerational persistence of human capital by 30 percent. We give suggestive evidence that these findings are explained by asymmetric impacts on parents' time investments owing to family disruptions and (lack of) substitutability of parents' time inputs.