BACKGROUND: This study reports the responses of patients with confirmed depressive illnesses to different treatments in the WHO Mental Disorders in General Health Care study, conducted in 15 cities around the world. AIM: To discover how depressions recognized by the doctor compare with unrecognized depressions, both in terms of the initial illnesses and their outcomes, and to compare the outcomes of those depressions treated with antidepressants with those treated with daytime sedatives. METHOD: The design of the study was naturalistic, in that physicians were free to treat patients however they wished. Patients with confirmed depressive illnesses were assigned to four groups: treatment with an antidepressant; treatment with a daytime sedative (usually a benzodiazepine); patients recognized as having depression by the physician but were not offered drug treatment; and patients unrecognized as having depression by their physician. RESULTS: Both groups receiving drugs had illnesses of equal severity, were demographically similar to one another, and had similar previous histories of depression. Those receiving antidepressants had significantly fewer overall symptoms and fewer suicidal thoughts than those treated with sedatives. By the end of one year, differences between the groups had disappeared: patients not given drugs had milder illnesses but did significantly better than those receiving drugs, both in terms of symptoms lost and their diagnostic status. Unrecognized depressions were less severe than recognized depressions, and had a similar course over the year. CONCLUSIONS: Patients receiving antidepressants were better in terms of overall symptoms and suicidal thoughts than those treated with sedatives at three months, but this advantage does not persist. Depression emerges as a chronic disorder at one-year follow-up--about 60% of those treated with drugs, and 50% of the milder depressions, still meet criteria for caseness. The study does not support the view that failure to recognize depression has serious adverse consequences, but, in view of the poor prognosis of depression, measures to improve compliance with treatment would appear to be indicated.