Long working hours and depressive symptoms: moderating effects of gender, socioeconomic status, and job resources Abstract Purpose Systematic reviews and meta-analyses have found inconsistent associations between working hours and depressive symptoms. The purpose of this study was to investigate the possible moderators of this association, using data from a large-scale cross-sectional survey. Methods A total of 16,136 Japanese employees (men 83.5%; women 16.5%) responded to a self-administered questionnaire inquiring about overtime working hours during the previous month and depressive symptoms (Center for Epidemiologic Studies Depression scale), as well as moderating factors including gender, age, marital status, socioeconomic status, commuting time, sleeping hours per day, job control and worksite social support (Job Content Questionnaire), neuroticism (Eysenck’s Personality Questionnaire Revised), and social desirability (Social Desirability Scale) (response rate, 85%). We conducted sequential regression analyses to investigate the main effects and interaction effects of all moderating variables. Results The association between overtime working hours and depressive symptoms was significantly moderated by gender (interaction effect: β = 0.03), age (β = − 0.02), manager (β = 0.03), sleeping hours (β = − 0.02), job control (β = − 0.03), and neuroticism (β = 0.02). Among workers engaged in 80 + hours of overtime, higher depressive symptoms were reported by women, younger employees, non-managers, employees with low job control, low worksite social support, and high neuroticism. A significant main effect of long overtime working hours on depressive symptoms was also observed even after controlling for all independent variables (β = 0.02). Conclusions Long overtime working hours is associated with depressive symptoms. We also found significant heterogeneity in the association according to employee characteristics, which may explain the inconsistent findings in previous literature.