AbstractBackground National trends in adolescentís marijuana risk perceptions are traditionally used as a predictor of concurrent and future trends in adolescent marijuana use. We test the validity of this practice during a time of rapid marijuana policy change. Methods Two repeated cross-sectional U.S. nationally-representative surveys of 8th, 10th, and 12th-graders: Monitoring the Future (MTF) (1991-2015; N?=?1,181,692) and National Survey on Drug Use and Health (NSDUH) (2002ñ2014; N?=?113,317). We examined trends in the year-to-year prevalence of perceiving no risk of harm in using marijuana regularly, and prevalence of regular marijuana use within the previous month. A piecewise linear regression model tested for a change in the relationship between trends. Similar analyses examined any past-month use and controlled for demographic characteristics. Results Among MTF 12th-graders, the prevalence of regular marijuana use and risk perceptions changed similarly between 1991 and 2006 but diverged sharply afterward. The prevalence of regular marijuana use increased by ~1 percentage point to 6.03% by 2015. In contrast, the proportion of 12th-graders that perceived marijuana as posing no risk increased over 11 percentage points to 21.39%. A similar divergence was found among NSDUH 12th-graders and other grades, for any past month marijuana use, and when controlling demographic characteristics. Conclusions An increase in adolescent marijuana use has not accompanied recent rapid decreases in marijuana risk perceptions. Policy makers may consider broader prevention strategies in addition to targeting marijuana risk perceptions. Further monitoring of predictors of marijuana use trends is needed as states legalize recreational marijuana use.