Background To clarify the direction of the association between frequency of cannabis use, depressive symptoms and suicidal ideation from 15 to 20 years using cross-lagged analyses. Method We included 1606 adolescents from the province of Québec followed since 1997 with information on frequency of cannabis use (none/monthly/weekly), depression (defined as being in the top 10% symptoms) and serious suicidal ideation at ages 15, 17 and 20 years. Results The prevalence of weekly cannabis use increased from 7.0% at age 15 years to 15.6% by age 20 years. Adolescents who reported using cannabis weekly at one age were 11 to 15 times more likely to continue using cannabis over time. In longitudinal cross-lagged analyses, weekly cannabis use at age 15 was associated with greater odds (OR=2.19, 95% CI=1.04-4.58) of suicidal ideation two years later. However, other substance use (alcohol, tobacco, other drugs) fully explained this association. Further, depression predicted subsequent weekly cannabis use, even after adjusting for comorbid other substance use (eg, for depression at 15 years predicting cannabis use at 17 years: OR=2.30, 95% CI=1.19-4.43). Limitations Quantity of cannabis consumed was not measured. Conclusion Findings suggest that depressive symptoms in adolescence may represent a risk factor for weekly cannabis consumption, which once initiated is likely to remain chronic. Weekly cannabis use increased risk for suicidal ideation, but not independently from other substance use including alcohol, tobacco and other drugs.< Réduire