Since citalopram was first approved in 1989, it has been prescribed to an estimated > 600 000 patients. An integrated safety database has been prepared, including data from 3107 patients from 24 clinical trials. In placebo-controlled trials, nausea, dry mouth, somnolence, increased sweating, tremor, diarrhoea, and ejaculation failure, mostly of mild to moderate severity, occurred significantly (p < 0.05) more frequently with citalopram. The excess incidence of these events over placebo was always less than 10%. In pooled comparative studies, citalopram's tolerability profile was similar to that of other selective serotonin reuptake inhibitors (SSRIs) and superior to that of tricyclic antidepressants (TCAs). Spontaneous adverse event reports arising from clinical use have confirmed the safety profile defined during the trials programme. Specific monitoring of all serious adverse events from around 10 000 patients receiving citalopram in clinical trials (including small open studies) has indicated a low potential for convulsions and extrapyramidal effects. There is no evidence of withdrawal phenomena on abrupt discontinuation, no clinically relevant effects on cardiac or laboratory parameters, and little or no effect on psychomotor function. When taken in overdose alone, citalopram appears to have a relatively wide margin of safety. Citalopram has been well tolerated in both short- and long-term use, and the profile seen in trials has been confirmed in the clinic.