As a test of the efficacy of desipramine (DMI) in the treatment of cocaine dependence, 59 cocaine-dependent males, maintained on methadone for the treatment of opiate dependence, completed a 12-week, random-assignment, placebo-controlled trial of this medication. At the end of treatment, there were no overall differences between the placebo and DMI groups on a range of outcome measures, including urine toxicology tests. However, an interaction between psychiatric diagnosis and outcome was seen when the sample was divided into those with (51%) and those without (49%) antisocial personality disorder (ASP). Patients with ASP made few gains with either DMI or placebo. Those without ASP made a number of gains with DMI but not placebo, particularly in the areas of psychiatric symptoms, legal status, and family problems. DMI had a significant effect on the psychiatric symptoms and personal adjustment problems, but not the cocaine use, of non-antisocial cocaine abusers. The negative influence of ASP that has been seen in studies of psychosocial therapies for substance-use disorders may also apply to pharmacological therapies.