Eighty women taking part in a population study were subjected to a dexamethasone suppression test (DST) intended as a diagnostic aid for melancholia. The women were selected systematically from two age strata, 38 and 50 years. Fifteen subjects (19 per cent) were found to be non-suppressors. High post-dexamethasone serum cortisol concentrations were not the result of elevated concentrations of the main cortisol binder, transcortin. There were no differences between suppressors and non-suppressors as regards depressive symptoms, strain experience, body mass, gynaecological history, drug use, smoking, erythrocyte sedimentation rate, number of leucocytes, activity of serum aminotransferases and gamma-glutamyltransferase, serum iron, bilirubin, ferritin content, serum growth hormone or serum prolactin. However, the nonsuppressors reported a significantly lower (P less than 0.01) orgasmic capacity in a questionnaire inquiry about two weeks before the DST. The outcome of the study indicates that DST as the presently recommended procedure for out-patients has a lower specificity for melancholia than has been reported previously.