Thirty-four patients with seasonal affective disorder, winter depression type (WD) were randomly distributed to receive the selective monoamine oxidase-A inhibitor moclobemide (400 mg daily) or placebo in a double-blind, parallel group study lasting for up to 14 weeks. Severity measures were the Montgomery-Asberg Depression Rating Scale (MADRS) extended with characteristic symptoms of WD; summed score of the "atypical" symptoms hypersomnia, hyperphagia and carbohydrate craving; and Clinical Global Impressions (CGI). After 3 weeks, patients with unsatisfactory response were switched to open moclobemide. Three patients on placebo dropped out before 3 weeks. Extended MADRS and CGI showed no significant difference between the groups at 3 weeks, whereas the atypical score was reduced significantly more on moclobemide than on placebo already after one week. Nonresponders after 3 weeks (9 of 16 on moclobemide and 7 of 15 on placebo) improved rapidly after being given open moclobemide. Predictor analysis showed a remarkably high negative correlation between improvement at 3 weeks (extended MADRS) and age in the placebo group and a strong, nonsignificant trend in the same direction in the moclobemide group. Dichotomizing the patients according to the median age (45 years) resulted in a somewhat better effect of moclobemide than placebo in the older age group. There were no significant differences in side effects between moclobemide and placebo.