Distinction between relapse and recurrence is a major difficulty in studies assessing the efficacy of long term antidepressant treatment. A four to six months symptom-free period is required to talk of recovery from depression and plan a preventive treatment, i.e. prevention of recurrence. Compounds are not selected according to their biochemical profile; the most recent antidepressant drugs were studied in that indication through long term controlled trials. In unipolar patients, recent studies favour the prescription of high doses of antidepressant drugs during at least five years. In bipolar patients, combination of long term antidepressant to lithium treatment may be of benefit to patients whose index episode was of depressive type, but appears useless in other cases. In all cases, joint psychotherapy and the quality of the relationship between the patient and the prescribing physician influence compliance and outcome. Dysthymia, panic disorder and obsessive-compulsive disorder constitute other acknowledged cases for long term prescription of antidepressants. These indications remain empirical till controlled trials have not specifically assessed each antidepressant compound.