In this paper we use a parental leave reform directed towards fathers to identify the causal effects of paternity leave on children's and parents' outcomes. We document that paternity leave causes fathers to become more important for children's cognitive skills. School performance at age 16 increases for children whose father is relatively higher educated than the mother. We find no evidence that fathers' earnings and work hours are affected by paternity leave. Contrary to expectation, mothers' labor market outcomes are adversely affected by paternity leave. Our findings do therefore not suggest that paternity leave shifts the gender balance at home in a way that increases mothers' time and/or effort spent at market work.