In this paper, we use data from a corruption survey conducted among top politicians and high level civil servants in 290 Swedish municipalities in 2007 to investigate the effects of government size on corruption. We construct several measures of corruption based on the survey, and combine these corruption measures with detailed administrative municipality level data to estimate the effect of local government resources on corruption. In cross-sectional analysis, we find a robust and negative association between total public expenditure and corruption. When we use lagged population growth rates and age structure as instruments for expenditure in 2-stage-least-squares regressions, the point estimates remain negative, but are no longer significant. In contrast with standard political economy models, where a bigger public sector is typically assumed to cause problems with corruption and public office abuse, our results suggest that corruption pressures may be particularly high when government resources are limited.