Two self-report experiments examined how religiosity affects attributions made for the outcome of a tornado. Undergraduate students and online adults read a fictional vignette about a tornado that hits a small town in the United States. The townspeople met at church and prayed or prepared emergency shelters for three days before the tornado; either no one died or over 200 people died from the tornado. Participants made attributions of cause to God, prayer, faith, and worship. In both studies, individuals identifying (...) as Christian made more attributions to God, prayer, faith, and worship, but only when no one died; when townspeople died, Christian participants made fewer attributions to God, prayer, faith, and worship. Individuals identifying as agnostic or atheist did not show this bias. Directions for future research in terms of implicit religious beliefs and normative evaluations of religion are discussed. (shrink)