We tested the efficacy of bupropion in the treatment of persons with pathological gambling (PG). Nondepressed, healthy subjects with Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition PG were randomly assigned to placebo or flexibly dosed bupropion in a 12-week double-blind trial. Outcome measures included the Yale-Brown Obsessive-Compulsive Scale modified for PG, the Gambling Severity Assessment Scale, the Clinical Global Impression Improvement and Severity Scales, the Global Assessment Scale, the Timeline Follow Back, the Attention-Deficit/Hyperactivity Disorder Rating Scale, and the Sheehan Disability Scale. Thirty-nine subjects (28 men, 11 women) were randomized to bupropion (n = 18) or placebo (n = 21). The 2 groups were similar on demographic and clinical measures. There were few differences between the treatment groups on any primary or secondary outcome measure, although subjects in each cell experienced significant improvement. Of subjects with at least 1 postrandomization visit, 35.7% of bupropion and 47.1% of placebo recipients experienced "much" or "very much" improvement on the Clinical Global Impression Improvement Scale. The trial was complicated by a high noncompletion rate (43.6%). Bupropion was well tolerated. Bupropion and placebo recipients did equally well in a short-term trial, with improvement seen as early as the first week of treatment. The high placebo response rate and the high noncompletion rate each reflect the challenge inherent in treating persons with PG.