This paper studies the effects of non-pharmaceutical interventions during the 1918 epidemic in Sweden, focusing on school closures. We combine several data sets with newly collected archival data on the timing of school closures to estimate the short- and long-run effect of school closures during the pandemic on mortality, education, earnings and employment. Evidence from a staggered roll-out event-study design shows more (influenza) deaths in the week leading up to the closure of the school and the four following weeks, relative to areas keeping schools open. These findings indicate that--on average--schools were closed in response to a worsening local epidemic, but were otherwise a successful policy in achieving local epidemic control. We also document that a relatively fast implementation reduced epidemic mortality rates. Our preliminary results suggest no long-run effects on mortality, education, and earnings. Joint paper with Christian M. Dahl, CasperW. Hansen, Peter S. Jensen, Martin Karlsson.