The goal of this study is to examine different criteria for the diagnosis of dysthymic disorder (DD). Using a two-stage design, records of 37 subjects diagnosed with DSM-III-R DD were systematically reviewed. Seven of these patients met criteria except for having depression less than 50% of the time. The initial evaluation involved a comparison of the diagnostic assessments using symptom criteria of DSM-III-R and ICD-10, proposed DSM-IV criteria for DD, DSM-III-R major depressive disorder (MD) criteria, and the Columbia criteria for atypical depression (AD). A subsample of 16 subjects who were assessed at least twice several months apart was then extracted. Agreement between the assessments with the proposed DSM-IV symptom list was determined. In the initial assessment, all 37 subjects (100%) met DSM-III-R and ICD-10 criteria, proposed DSM-IV criteria for DD, and DSM-III-R MD criteria, with percentages of mean positive items on the four criteria being 77%, 84%, 89%, and 77%, respectively. In addition, nine subjects (24.3%) also met diagnostic criteria for AD. All 16 subjects (100%) of the subsample again met proposed DSM-IV criteria for DD when assessed 75 to 145 days after their initial assessment. Items with the highest agreement between each two assessments were pessimism and low self-esteem, whereas social withdrawal and decreased activity had lower agreement. The proposed DSM-IV criteria for DD had good sensitivity and reliability in assessing core symptoms of DD. The use of different criteria for DD is discussed.