Patients with a depressive illness with 4 major symptoms of depression and a score of at least 17 on the Hamilton Depression Scale (1-17) (HDS) were allocated to a randomized double-blind group comparative study in general practice. After retrospective analysis, all 81 patients except one were characterized as suffering from a 'Definite Major Depressive Disorder', as defined by Spitzer et al. (1978). After 6 weeks of treatment with a daily dosage of 600 mg femoxetine or 150 mg amitriptyline, no statistically significant differences between the 2 treatment groups were observed, either when using the HDS or the clinical global assessment scale. Confidence limits of 95% for differences between therapeutic effect showed a non-significant tendency in favour of amitriptyline. During treatment, there were statistically significant differences in the reduction of HDS score between the 2 treatments in week 2. These differences were the result of amitriptyline's significantly greater effect on the 3 sleep items at week 2, as indicated by the results of single item analysis. Drop out rates due to side effects were between 14-15% in both treatment groups. Of the patients treated with femoxetine, 38% experienced no side effects, compared to 14% of patients treated with amitriptyline. Nausea was the side effect most commonly reported by patients treated with femoxetine, whereas a significantly greater frequency of anticholinergic side effects was recorded during treatment with amitriptyline (P less than 0.05). Unlike amitriptyline, femoxetine did not increase body weight. Treatment with the active drug was continued after the trial period in 14 and 18 patients in the femoxetine and amitriptyline groups respectively.