Reducing youth cannabis use in Canada is a public health priority with schools of interest as a potential modifier of behavior and as a venue for prevention programming. This work aimed to provide a basis for future policy and programming by evaluating pre-legalization cannabis use change patterns in schools and the impact of school characteristics on these patterns. Average rates of cannabis use behavior change (initiation, escalation, reduction, cessation) were collected from 88 high schools located in Ontario and Alberta, Canada participating in the COMPASS prospective cohort study. There was little variability in cannabis use behaviors between schools with intra-class correlation coefficients lowest for cessation (0.02) and escalation (0.02) followed by initiation (0.03) and reduction (0.05). Modest differences were found based on school province, urbanicity and student-peer use. Cannabis ease of access rates had no significant effect. Fewer than half the schools reported offering school drug use prevention programs; these were not significantly associated with student cannabis use behaviors. In conclusion, current school-based cannabis prevention efforts do not appear sufficiently effective. Comprehensive implementation of universal prevention programs may reduce cannabis harms. Some factors (urbanicity, peer use rates) may indicate which schools to prioritize.