Objective: The aim of this study was to examine how reasons for substance use at age 18 relate to alcohol and marijuana use at ages 18 and 35 and to symptoms of alcohol use disorder and marijuana use disorder at age 35. Method: Bivariate correlation and multivariate regression analyses were conducted to examine the prediction of substance use and misuse by social/recreational, coping with negative affect, compulsive, and drug effect reasons for alcohol and marijuana use. Control variables included gender, race/ethnicity, parent education, and previous substance use (for age 35 outcomes). Results: Social/recreational, coping, and drug effect reasons for drinking predicted symptoms of alcohol use disorder 17 years later. Reasons for marijuana use were generally associated only with concurrent marijuana use; an exception was that drug effect reasons predicted marijuana use disorder at age 35. Conclusions: The long-term longitudinal predictive power of reasons for alcohol use (and, to a lesser extent, for marijuana use) suggests that adolescents' self-reported reasons, in particular those involving regulating emotions and experiences, may be early risk factors for continued use and misuse of substances into adulthood.