Antidepressants increase melatonin levels, but it is still unclear whether this effect is related to the improvement of depressive symptoms or to unrelated pharmacological action of antidepressants. To answer this question, the effect of antidepressants on 6-sulphatoxymelatonin (aMT6s), the main melatonin urinary metabolite, was examined in drug-free depressed patients - most of them antidepressant-naive. aMT6s was evaluated in 34 depressed patients, before and after 8 weeks of placebo (n = 12) or antidepressant (n = 22; fluoxetine, duloxetine or Hypericum perforatum). Both groups showed an improvement of depressive symptoms after treatment compared to baseline (Hamilton Depression scores): 17.0 +/- 1.4 vs. 9.0 +/- 2.8, P = 0.007 for placebo, and 18.6 +/- 1.1 vs. 11.8 +/- 1.6, P < 0.001 for antidepressants). After treatment, aMT6s levels increased after antidepressants (P < 0.01), but not after placebo (P > 0.05). As depressive symptoms improved both in patients taking antidepressant and in those taking placebo, but an effect of antidepressants could only be seen in those taking antidepressants, we suggest that melatonin changes after antidepressants are more likely due to a pharmacological action of these drugs on melatonin secretion.