The present study examined putative environmental predictors of adolescent substance use, using a prospective adoption design to distinguish between environmental mediation (i.e., parenting influencing adolescent substance use), passive gene–environment correlation (i.e., parental genetic predisposition influencing the association between parenting characteristics and adolescent substance use), and evocative gene–environment correlation (i.e., children’s genetic predisposition influencing parenting). Longitudinal data from the Colorado Adoption Project (395 adoptees, 491 nonadoptees, 485 adoptive parents, and 490 biological parents) were examined. Children (48% girls) were assessed at ages 1 to 17 years. Over 90% of the sample were non-Hispanic White. Associations between parenting and adolescent substance use were compared between adoptive and nonadoptive families. Positive, negative, and inconsistent parenting measures in early childhood through adolescence were not consistently associated with adolescent substance use, with only 6% of correlations being statistically significant (r = −0.152 to .207). However, parent–child relationship quality assessed from childhood to adolescence and orientation to parents assessed during adolescence were significantly, negatively associated with adolescent substance use, with 71% of correlations being statistically significant (r = −0.88 to −0.11). There was little evidence of sex differences in the associations. Environmental mediation, rather than passive or evocative gene–environment correlation, explained most associations.