Drug surveillance data can be obtained from different sources: spontaneous French and European reporting, data from WHO, bibliographic analysis. We are more interested, in this paper, in the type of data obtained than in the surveillance of each antidepressant. Our study is focused on psychiatric and neurobehavioural effects of antidepressants. The French drug surveillance database was examined for reactions associated with fluoxetine. Psychiatric side effects are not the most frequent. The psychiatric safety profiles of the three SSRIs (fluoxetine, fluvoxamine and paroxetine) are similar. However withdrawal reactions with fluvoxamine and paroxetine occur in a greater proportion of reports (13 and 14 per cent) than with fluoxetine (1.5 per cent). In contrast, pharmacodependance was observed in 79 per cent of reports with amineptine. Tricyclic antidepressants do not seem to confer increased risk of teratogenesis. Preliminary data regarding risk of prenatal exposure to fluoxetine suggest that its use during pregnancy is relatively safe. Data regarding neurobehavioural effects of prenatal exposure are lacking for all antidepressants. Cognitive disorders induced by antidepressants are complex, due to the involvement of several factors that can intervene in the pathogenesis and evaluation of these disorders : most studies evaluate the modifications of neurobehavioural effects in healthy subjects, few studies concern chronic patients. Proposals are made to improve the evaluation of these side effects.