OBJECTIVE The aim of this study is to investigate whether depressive (DT), hyper thymic (HT), cyclothymic (CT), irritable and anxious temperaments as identified by Turkish version of Temperament Evaluation of Memphis, Pisa, Paris and San Diego Auto questionnaire (TEMPS-A) are characteristic to depressive or bipolar disorder (BD) and to compare underlying affective temperamental differences. METHODS 68 patients with recurrent major depression (MD-R), 50 patients with single episode major depressive disorder (MD-S), 84 euthymic bipolar patients and 100 healthy controls were included in the study and evaluated with TEMPS-A. RESULTS There was a gender difference between patient groups. The females were more dominant in the MD-R group. The mean age of onset of illness of bipolar patients was lower than the depressive patients and the mean duration of illness of bipolar patients was longer than those of the MD-R patients. The prevalence of the rate of any dominant affective temperament was significantly higher in the MD-R group than other study groups. The prevalence rate and scores of DT were significantly higher both in MD-R and MD-S groups (25.0% and 12.0% respectively) than the BD group (2.4%) and control subjects (3.0%). The prevalence rate and scores of CT were higher among the patients groups than the controls. None of the subjects except BD patients had HT. CONCLUSIONS With the limitations of the study, it is thus reasonable to speculate that affective temperament, to a degree, determine the nature and the existence of the mood disorders.