Religious people are thought to be more prosocial than nonreligious people. Laboratory studies of this using ultimatum, dictator, public goods and trust games have produced mixed results, which could be due to lack of context. This article examines the relationship between religion and prosocial behaviour using data from a context-rich, naturally occurring field experiment that closely resembles the dictator game - tipping in restaurants. Customers were surveyed as they left a set of restaurants in Richmond, Virginia, in the summers of 2002 and 2003. Our findings reveal no evidence of religious prosociality.