Whether monetary incentives to change behavior work and how they should be structured are fundamental economic questions. We overcome typical data limitations in a large-scale field experiment on vaccination (N = 5,324) with a unique combination of administrative and survey data. We find that guaranteed incentives of $20 increase uptake by 13 percentage points in the short run and 9 in the long run. Guaranteed incentives are more effective than lottery-based, prosocial, or individually-targeted incentives, though all boost vaccinations. There are no unintended consequences on future vaccination or heterogeneities based on vaccination attitudes and incentivized economic preferences. Further, administrative data on relatives shows substantial positive spillovers. Our findings demonstrate the great potential of incentives for improving public health and provide guidance on their design.