This article delineates the conceptual models used when medications are prescribed for patients with personality disorders and reviews the data on the efficacy of these medications. Studies before 1980 are difficult to interpret because of changes in diagnostic criteria. Nonetheless, early studies on non-DSM-III disorders such as pseudoneurotic schizophrenia, emotionally unstable character disorder, hysteroid dysphoria, and subaffective disorders indicated the potential utility of pharmacotherapy for treating personality disorders. Models to consider in evaluating the possible use of medications for treating personality disorders are: (1) treating the disorder itself; (2) treating symptom clusters within and across disorders; and (3) treating associated axis I disorders. Among the current personality disorders, borderline personality disorder has been the most extensively studied, with antipsychotic agents being the most well-documented treatment. Monoamine oxidase inhibitors, fluoxetine, and carbamazepine show promise. Schizotypal disorders may respond to low-dose antipsychotic drugs. Although heuristically valuable, the symptom cluster approach to treatment has not yet been validated. Axis I disorders, especially depression, are frequently associated with all personality disorders. Dependent personality disorder is linked to panic disorder with agoraphobia, whereas avoidant personality disorder is associated with social phobia and panic. In general, pharmacotherapy for axis I disorders is less effective in the presence of a comorbid personality disorder. Despite the modest benefits seen in many studies, pharmacotherapy can add significantly to the overall treatment of those with personality disorders. Future research must carefully assess the effect of comorbid axis I disorders on responses. The symptom cluster/psychobiologic dimension approach should be investigated in clinical studies.