In recent years there has been a change in the risk-benefit ratio of antidepressants characterized by improvement in terms of side effects with second generation drugs and efficacy in new indications (anxiety disorders for example). However, no significant gain in terms of efficacy in major depressive disorders has been obtained. Some aspects of the risk-benefit ratio of antidepressants should be better assessed before being optimized: for example, onset of action, efficacy and safety of long term treatment, in particular dose and duration of treatment, suicidal thoughts and suicide in treated patients, effects of antidepressants in bipolar disorders, compliance and economic evaluation. In conclusion, the evaluation of risk-benefit ratio associated with antidepressants, drugs of proven efficacy in many indications, should take into account the risks of not treated depressive disorders (chronicity, social functioning and mortality).