Background: Dysthymia is a highly prevalent disorder with rates varying from 3 to 6%. It is estimated that 7% of primary care and up to 36% of the psychiatric outpatients have this disorder. Dysthymia is characterized by a pattern of low intensity and long duration depression, leading to impairment in many aspects of the patient's life (work, family). However, it is a condition which is not frequently diagnosed, thereby remaining untreated and leading to chronification, risk of developing depression and substance abuse. Despite the well-known fact that dysthymia is underdiagnosed, there are no studies in Brazil regarding this problem. This paper presents a study on the prevalence of dysthymia in an outpatient psychiatric population in Sao Paulo, Brazil, identifying factors associated with the underdiagnosis of dysthymia. Methods: Two raters carried out a study of 81 patients from a university outpatient clinic in Sao Paulo, using the SCID-I/P. The K coefficient was used to measure the inter-rater reliability for the dysthymia diagnosis. The SCID-I/P diagnosis was compared to consulting psychiatrist's diagnosis. Factors associated with underdiagnosis were identified. Results and conclusions: 27% of the sample had SCID dysthymia. They presented more comorbidities than non-dysthymics. Only 27% of these SCID dysthymics were considered 'pure' dysthymics. Clinicians had identified only 37% of these SCID diagnosed cases as dysthymic. The main reason for missing diagnosis of dysthymia in this sample was axis I comorbidity. Limitation: As we did not investigate the presence of personality disorder, the impact of axis II on misdiagnosis of dysthymia could not be investigated in our sample. (C) 2002 Elsevier Science B.V. All rights reserved.