Aim. The study aimed to analyse the bipolarity features between patients with treatment-resistant depression and patients with remission of a depressive episode. Methods. The population studied included patients aged 18-65 who had a diagnosis of first or recurrent major depressive disorder. Each psychiatrist included 5 consecutive patients, from all the patients he is treating, where the depressive episode was assessed as treatment-resistant (TR) and 5 consecutive patients with a depressive episode in remission (RE). One thousand and fifty-one patients entered the study: 570 in the treatment-resistant depression group and 481 in the remission group. All patients were assessed using the Mood Disorder Questionnaire (MDQ) and Hypomania Checklist (HCL-32). The assessment of current depressive symptoms was done by means of a 17-item Hamilton Depression Rating Scale. Results. The percentage of patients screened positive by MDQ was significantly higher in TR than in RE (13.7 vs. 5.6% p<0.001). Also, the percentage of patients having positive response to 14 or more items of the HCL-32 was significantly higher in TR than in RE (43.9% vs. 30.0%, respectively, p<0.001). The multiple logistic regression model showed that independent risk factors for treatment resistance were: bipolar features in HCL-32 scale, scoring 6 or higher on MDQ, age at first onset <= 20 years, more than three previous depressive episodes and partial remission after the previous depressive episode. Conclusions. Identified independent risk factors for treatment resistance suggest possible association between bipolarity and worse response to antidepressant drugs in patients with mood disorders.