Hypericum extract (HE) might be favourably active in depressed patients with reversed vegetative signs (RVS). Therefore, we performed an exploratory subgroup analysis of a three-armed study to compare HE, fluoxetine, and placebo in patients with major depressive disorder (MDD) in a 12 wk trial. A total of 135 patients were randomized to 12 wk treatment with HE LI 160 (900 mg/d), fluoxetine (20 mg/d), or placebo. Patients with RVS were defined in two steps, according to DSM-IV. First, patients with melancholy-related vegetative signs were excluded. Secondly, patients had to have at least one score of 2 for the items 22-26 of the HAMD-28 scale, which are related to hypersomnia and hyperphagia. Twenty-seven patients remained in the group. Analysis of covariance (ANCOVA) was applied using the HAMD-17 score. Secondly a chi2 test for response was performed, using the same and further an adapted criterium as in recently published studies. ANCOVA revealed a trend to a global difference. Post-hoc analysis showed a trend to superiority of HE compared to placebo and to fluoxetine, but a very large effect size for both differences. Fluoxetine was not different from placebo. The adapted response criterium showed a significant global difference as well as a significant superiority of HE over placebo and over fluoxetine. These data are based on a small sample size and must be considered tentative. A characterization of vegetative features of patients with depression could lead to an overall increased effect size in the treatment with HE.