To what extent do SSRIs increase the risk of suicide or violent behavior? Recent data indicate that some SSRIs lack a favorable 'risk-benefit' profile for treating childhood and adolescent depression. In addition, certain recent documents have cast doubt on previous conclusions dissociating antidepressant from violence. Beyond the debate on the need for improved transparency in clinical studies, this controversy raises other critical issues of SSRI use that merit our consideration. Firstly, is there an overuse of antidepressants, and does the risk of suicide increase with the prescription of antidepressants in the whole population? Secondly, is the response to antidepressants modified during adolescence? If so, by what mechanisms? We know that SSRI treatment can trigger mood changes in undiagnosed young bipolar patients resulting in agitation and disinhibition, and such effects may lead to suicide and violence. Thirdly, recent extensive literature has shown that some cases of depression are worsened by antidepressants and new data has suggested that bipolar depression can be improved by atypical antipsychotics. Finally, current criteria defining the depressive state are very similar to those employed during the 19th century, a pre-therapeutic period. Currently, while it is recognised that depressive mood is a very broad construct, there is still only one definition in international classifications to describe a major depressive episode. There is a clear need to develop an evidence-based approach to psychiatry aiming at delineating diagnostic categories predictive of the response to treatments.