Several prescription database studies suggest major differences between antidepressants in the proportion of patients in general practice treated with doses likely to elicit a true drug response. However, in these studies it has been difficult to differentiate prescriptions for depression from those for other indications, or to distinguish lower starting doses from the final treatment doses. We investigated possible differences between types of antidepressant in the proportions of patients receiving adequate treatment doses for depression from the primary health care services of Helsinki. Doctors at 22 (71%) of the 31 health centres in Helsinki were surveyed with regard to their antidepressant prescriptions over a period of 2 working weeks. There were marked differences in dosaging adequacy between the various types of antidepressant prescribed for depression at final treatment doses. Overall, 71% of the prescriptions for the tricyclic antidepressants, but only 13% of those for the other antidepressants, mainly selective serotonin reuptake inhibitors, were for low doses that are generally considered to be ineffective (P<0.001). These findings endorse the emerging perception that, in general practice, tricyclic antidepressants are usually prescribed for depression in too low doses, and that a greater reliance on other antidepressants would probably improve the effectiveness of treatment in primary health care.