This hypothesis asserts that involvement in deviant behavior is inversely related to religiosity. Early revisions of this hypothesis stressed the importance of religiosity on violations of ascetic norms over violations of secular norms. More recent revisions have stressed the interactive effects of religiosity and contextual factors such as denominational norms and aggregate religiosity. Each of these respecifications of the original Hellfire hypothesis is evaluated here with survey data on self-reported alcohol and marijuana use from a sample of 3,065 male and female adolescents in grades seven through twelve in three midwestern states. Results suggest that most of these revisions are only marginally more useful than the original formulation for explaining adolescent alcohol consumption and are largely irrelevant with regard to marijuana use. We find that the more parsimonious proposition of a direct religiosity effect alone does about as well in explaining alcohol and marijuana use among adolescents as the more complex contextual propositions. (Author abstract) 7 tables, 4 notes, and 88 references.