Objectives: A casual relationship between temperament, job stress and depressive symptoms has not been established yet. The purpose of this study was to assess the relationships between job stress, temperament and depressive symptoms in female nurses at a Japanese general hospital. Material and Methods: A self-report survey was conducted among 706 nurses. We measured job stress, temperament, and depressive symptoms using the Brief-Job Stress Questionnaire, the TEMPS-A and a screening scale of items from the Ministry of Health, Labour and Welfare of Japan. In order to examine the causal relationship between the measures the stepwise multiple regression and path analyses were used. Results: Depressive symptoms were modestly correlated with job stress (γ = -0.23-0.30). Except for hyperthymic temperament measures, the correlations between depressive symptoms and temperament types were significant and moderate (γ = 0.36-0.50). Overtime, job control as well as depressive and cyclothymic types of temperament were significantly correlated with depressive symptoms (β = 0.15, p < 0.05; β = 0.19, p < 0.01; β = 0.26, p < 0.001; β = 0.32, p < 0.001, respectively). Path-analysis revealed that depressive and cyclothymic types of temperament influenced depressive symptoms both directly (β = 0.67, p < 0.001) and indirectly via job stress (β = 0.35, p < 0.001 from temperament to job stress; β = 0.20, p < 0.05 from job stress to depressive symptoms). Irritable and anxious types of temperament and quantitative job overload did not contri­bute to the path-analytic model. Conclusions: Health care professionals should consider temperament, especially depressive and cyclothymic types, in order to help employees cope better with job stress factors. We need further research about the effective intervention to help employees better cope with their job stress.