Aim. This is a pilot study of the multicenter TRES-DEP (Treatment Depression) Project. Its' aim was the detection and analysis of bipolar spectrum features in drug resistant unipolar depression patients in comparison to depressive patients who responded to standard antidepressant treatment and remitted. Sociodemographic and clinical parameters of the patients were also described. Method. 50 drug-resistant (group 1) and 50 non-drug resistant (group 2) 18-65 year old patients, fulfilling ICD-10 criteria for depressive episode or recurrent depressive disorder were included in the study. The main exclusion criteria were: > 18 scores in the 17-item Hamilton Depression Rating Scale, treatment with mood stabilizers, the diagnosis of substance misuse. dementia, severe neurological or other somatic disease. The presence of bipolar spectrum was detected by Mood Disorder Questionnaire (MDQ) and Hypomania Checklist Scale (HCL). Results. There was statistically significant more patients fulfilling MDQ or HCL bipolar spectrum criteria in group 1 than in group 2 (44% vs. 12%, p < 0.001 and 62%% vs. 34%, p<0.005 respectively,). Significantly more drug resistant patients comparing to non-refractory patients, considered their last remission as partial (88% vs. 52%, p=0.001). Non-refractory patients had a history of fewer depressive episodes (5.1 +/- 3.8 vs. 8.5 +/- 5.0; p=0.001) and reported a longer time from the last hospitalization (41.9 +/- 71.1 vs. 14.8 +/- 26.5 months, p<0.005). Significantly more patients from group I reported a family history mental disorders, especially of alcoholism. Conclusion. The results of the study suggest that misdiagnosed and inadequately treated bipolarity may be one of the main reasons of non-response in the treatment of depression.