We argue that insurance against idiosyncratic risk among members of religious communities is more valuable when members face greater common background risk. We test this hypothesis by examining whether religious communities in the nineteenth-century US were larger in counties subject to greater temporal rainfall variability — a source of aggregate economic risk at a time when rain fed agriculture was the dominant economic activity in most counties. Our results indicate that religious communities had more members and greater seating capacity in counties with greater rainfall risk. This link is stronger among more agricultural counties or counties with low population densities. We also find the link between rainfall risk and religious membership to be stronger for rainfall risk during the agricultural growing season. Joint with Philipp Ager.