A growing empirical literature suggests that Protestant missions increase community cohesion. Yet, causally identified evidence is scarce, and the mechanisms underlying the link have received little empirical scrutiny. This paper explores two mechanisms through which Protestant missions may affect community cohesion: pro-social preferences and social networks. One the one hand, Protestant missions teach charity and establish regular venues of social interaction. On the other hand, Protestant missions propagate individual salvation and provide an identity along which communities may separate. The effect of Protestant missions on community cohesion is, hence, unclear. To make headway on these conflicting predictions, I take advantage of exogenous variation in missionary activity in southeastern Peru. I document that villages exposed to Protestant missions have lower levels of community cohesion compared to control villages. Using a causal mediation framework, I find sparser networks to predominantly mediate this reduction. Importantly, the reduction in community cohesion is also detectable within the missionary communities, which calls a club model of religion into question.