The rationale for pharmacological treatment of bulimia nervosa is summarized and a review of controlled therapeutic trials shows contradictory results. A number of antidepressant agents (tricyclics: imipramine, desipramine, amitriptyline; IMAO: phenelzine, isocarboxazide; trazodone; fluoxetine) appear more effective than placebo in double-blind controlled trials of 6 to 16 weeks. In similar studies, other antidepressants (mianserine, fluvoxamine) are ineffective. Improvement reported is often incomplete and the low percentage of patients totally abstinent at the end of treatment appears of poor pronostic value for long-term outcome. Methodological limitations of existing studies are discussed, and some psychopathological factors to consider in the assessment of therapeutic response are proposed.