Background: The endocannabinoid system (ECS) is increasingly being recognized as key regulatory system coupled with the glucocorticoid system implicated in the pathophysiology of major depressive disorder (MDD). However, prior studies examining the ECS in MDD have been inconclusive, of small sample size or of cross-sectional nature limiting interpretation of causal inferences or time-dependent effects. Methods: In a prospective community-based cohort study including 128 individuals (women: 108), depressive symptoms (PHQ-9) as well as hair cortisol and endocannabinoids were measured annually over four years (T1-T4). Cortisol, N-arachidonoylethanolamine (AEA), and 2-arachidonoyl-sn-glycerol/1-arachidonoyl-sn-glycerol (2-AG/1-AG) were extracted from 3 cm hair segments reflecting cumulative concentrations of the last three months prior sampling. Results: Cross-sectional group comparisons at baseline revealed reduced AEA and cortisol levels in the group with a positive MDD screening compared to individuals with low depressive symptomatology (both p <.05). Cross-lagged panel models showed that AEA levels at T2 were negatively associated with depressive symptoms at T3 (p <.05). Also, depressive symptoms at T3 were negatively associated with AEA levels at T4 (p <.01). The direction of association was reversed for 2-AG/1-AG, as 2-AG/1-AG levels at T1 were positively associated with depressive symptoms at T2 (p <.01). Conclusions: While cross-sectional analyses suggest higher depressive symptomatology to be associated with reduced AEA and cortisol release, longitudinal analyses reveal that primarily AEA levels are negatively associated with depressive symptoms. These longitudinal associations elucidate time-dependent relationships between depressive symptomatology and the ECS and further highlight AEA as potential treatment target in MDD.