Twenty variables, representing competing theories, were compared in their ability to explain church membership rates (CMRs) in 288 metropolitan areas for 1980, using multiple regression. Strong support emerged for social bond theory, which predicts low rates of church membership in communities with high social instability, such as that caused by geographic migration. Secularization theories, however, did not fare well. Their prediction that CMRs would be low in big cities was contradicted, and CMRs correlated positively with percent urban, rather than negatively. Although education correlates negatively with CMRs, this may be the effect of economic prosperity rather than the secularizing effect of education itself. Life cycle theories of church membership received only weak support.