We examine the effectiveness of church service bans in containing the spread of Covid-19 in Germany. We furthermore investigate how differences in the local religious affiliations affect infections and the effectiveness of church bans and other church-related restrictions. We find that, without a ban, infections per capita are higher in districts (Landkreise) with larger shares of religious population. In panel analysis, controlling for district fixed effects and a host of potential confounders, we find that church bans effectively reduce infections. For a ban in place for 14 days before a considered day, the predicted growth factor of infections is lower by 0.9 of its standard deviation. Finally, we show that Easter contributed significantly to the growth of infections in 2020 and 2021. The growth factor of infections was lower in regions with larger shares of Catholics and Protestants during Easter 2020 (when a church ban was in place) but not in 2021 (without a ban).