Evidence regarding the association between cannabis use and depression remain conflicting, especially as studies have not typically adopted a longitudinal design with a follow-up period that was long enough to adequately cover the risk period for onset of depression. Males from the Cambridge Study in Delinquent Development (CSDD) (N = 285) were assessed seven times from age 8 to 48 years to prospectively investigate the association between cannabis use and risk of major depressive disorder (MDD). A combination of multiple analyses (logistic regression, Cox regression, fixed-effects analysis) was employed to explore the strength and direction of effect within different developmental stages. Multiple regression analyses revealed that early-onset cannabis use (before age 18) but not late-onset cannabis use (after age 27) was associated with a higher risk and shorter time until a subsequent MDD diagnosis. This effect was present in high-frequency [(odds ratio (OR) 8.83, 95% confidence interval (CI) 1.29-70.79]; [hazard ratio (HR) 8.69, 95% CI 2.07-36.52)] and low-frequency early-onset users (OR 2.41, 95% CI 1.22-4.76; HR 2.09, 95% CI 1.16-3.74). Effect of increased frequency of cannabis use on increased risk of subsequent MDD was observed only for use during adolescence (age 14-18) but not at later life stages, while controlling for observed and non-unobserved time-invariant factors. Conversely, MDD in adulthood (age 18-32) was linked to a reduction in subsequent cannabis use (age 32-48). The present findings provide evidence implicating frequent cannabis use during adolescence as a risk factor for later life depression. Future studies should further examine causality of effects in larger samples.