Marijuana use on college campuses is prevalent and associated with high rates of abuse and dependence. The Marijuana Decisional Balance (MDB) scales measure perceived pros and cons toward marijuana use. Evidence supports reliability and concurrent validity of these scales, but the predictive validity has not yet been assessed. The current study evaluated the prospective predictive validity of pros and cons scales for marijuana use, as well as explored predictive validity for marijuana problem indicators. Secondary analyses included test–retest reliability and internal consistency, to provide additional evidence of psychometric properties. A total of 149 college students (57% recent marijuana users, 77% lifetime users) participated in a baseline survey, then completed a second survey one month later. All provided data on marijuana pros and cons, as well as use status in the past month. Users at each time point reported on use frequency, problems, and disorder symptoms. In the month between assessments, 55% of the students used marijuana. Both pros and cons subscales prospectively predicted use status in the subsequent month, but not use frequency. Pros prospectively predicted marijuana problems and dependence symptoms at follow-up, and remained a significant predictor of later dependence symptoms even after controlling for baseline dependence symptoms. In contrast, pros only marginally predicted abuse. Cons did not predict problems, abuse, or dependence symptoms. Pros and cons showed strong test–retest reliability (rs = 0.80–0.85) and internal consistency (alphas = 0.92–0.95). In a college sample, pros and cons of marijuana use demonstrated stability over one month, and prospectively predicted use. Pros may also have utility in predicting problems and dependence potential on college campuses.