Despite public sentiment to the contrary, recreational marijuana use is deleterious to adolescent health and development. Prospective studies of marijuana use trajectories and their predictors are needed to differentiate risk profiles and inform intervention strategies. Using data on 15,960 participants in the National Longitudinal Study of Adolescent to Adult Health, variable-centered approaches were used to examine the impact of childhood polyvictimization on marijuana onset, marijuana use from age 15 to 24 years, and marijuana dependence symptoms. Zero-Inflated Poisson latent class growth analysis (ZIP-LCGA) was used to identify marijuana use subgroups, and their associations with childhood polyvictimization were tested via multinomial logit regression within ZIP-LCGA. Results showed that the overall probability and frequency of marijuana use increased throughout adolescence, peaked in early adulthood, and diminished gradually thereafter. Polyvictimization was associated with earlier onset and greater overall use, frequency of use, and dependence symptoms. ZIP-LCGA uncovered four subgroups, including non-users and three classes of users: adolescence-limited users, escalators, and chronic users. Polyvictimization distinguished non-users from all classes of marijuana users. The findings underscore the lasting developmental implications of significant childhood trauma. Children who experience polyvictimization represent a group that may benefit from selective interventions aimed at preventing early, frequent, chronic, and dependent marijuana use.