Background The authors sought to model the different trajectories of psychotic‐like experiences (PLE) during adolescence and to examine whether the longitudinal relationship between cannabis use and PLE is mediated by changes in cognitive development and/or change in anxiety or depression symptoms. Methods A total of 2,566 youths were assessed every year for 4‐years (from 13‐ to 16‐years of age) on clinical, substance use and cognitive development outcomes. Latent class growth models identified three trajectories of PLE: low decreasing (83.9%), high decreasing (7.9%), and moderate increasing class (8.2%). We conducted logistic regressions to investigate whether baseline levels and growth in cannabis use were associated with PLE trajectory membership. Then, we examined the effects of potential mediators (growth in cognition and anxiety/depression) on the relationship between growth in cannabis use and PLE trajectory. Results A steeper growth in cannabis use from 13‐ to 16‐years was associated with a higher likelihood of being assigned to the moderate increasing trajectory of PLE [odds ratio, 2.59; 95% confidence interval (CI), 1.11–6.03], when controlling for cumulative cigarette use. Growth in depression symptoms, not anxiety or change in cognitive functioning, mediated the relationship between growth in cannabis use and the PLE moderate increasing group (indirect effect: 0.07; 95% CI, 0.03–0.11). Conclusions Depression symptoms partially mediated the longitudinal link between cannabis use and PLE in adolescents, suggesting that there may be a preventative effect to be gained from targeting depression symptoms, in addition to attempting to prevent cannabis use in youth presenting increasing psychotic experiences.