This article addresses two inconsistent findings in the literature on adolescent religious activity (RA) and substance use: whether a dose-response relationship characterizes the association of these variables, and whether the association varies by grade, gender, ethnicity, family structure, school type, and type of substance. Multinomial logistic regression analyses of a large, diverse data set of high school students in metropolitan Columbus, Ohio (n = 33,007), found marked differences in alcohol, marijuana, and cigarette use among youths who never, occasionally, or regularly participated in RA. Weekly RA was consistently associated with less substance use, yet occasional RA sometimes was associated with greater use. Four groups accounted for variations in the RA-substance use relationship: African American youths, younger White youths, 12th-grade White males, and 12th-grade White females. Researchers should avoid assuming the RA-substance use relationship is dose-response and consider the implications of this complexity for theory and practice. (Contains 2 tables and 3 figures.)