Background. The descriptive validity of the melancholic features specifier of the DSM-IV major depressive disorder (MDD) is uncertain. Little is known about its relationship to psychiatric co-morbidity, stability across episodes, or strength in predicting course of illness.

Method. The Vantaa Depression Study (VDS) is a prospective, naturalistic cohort study of 269 patients with a new episode of DSM-IV MDD who were interviewed with SCAN and SCID-II between 1 February 1997 and 31 May 1998, and again at 6 and 18 months. Ninety-seven (36%) MDD patients met DSM-IV criteria for the melancholic features specifier, and were contrasted with 172 (64%) subjects with a non-melancholic MDD. The duration of the index episode was examined using a life chart.

Results. We found no difference in rates of any current co-morbid Axis I or II disorders between melancholic and non-melancholic depressed patients. Of those who had melancholic features at the index episode and subsequent episodes during the 18-month follow-up, only 22% (5/23) presented melancholic features during the latter. The non-melancholic subtype switched to melancholic in 25% (8/32) of cases. Differences in the course of melancholic and non-melancholic depression were very minor.

Conclusions. The descriptive validity of the DSM-IV melancholic features specifier may be questionable in MDD. There appear to be no major differences in current co-morbidity, or course of depression between melancholic and non-melancholic patients. The consistency of DSM-IV melancholic features across episodes appears weak.