Compulsivity in alcohol-dependent patients is a frequent cause of early relapse in the post-detoxification period. The present study is a 2-month trial on detoxified alcoholics undergoing a double-blind placebo-controlled treatment with fluoxetine (20 mg/day). The rating instruments were the Hamilton Depression and Anxiety Scales, a visual analogue scale for alcohol craving and an original scale for evaluating alcohol withdrawal. The abstinence rate for fluoxetine-treated patients was significantly higher than in the placebo group, whereas no difference between treatments was found on the rating scales. Medical problems, additional psychiatric diagnoses, and family alcoholism were negatively correlated with abstinence. Two subgroups of patients having significantly different characteristics were identified as to the outcome, by means of cluster analysis. They are likely to represent two different stages in the evolution of alcoholism. Our results show that, independently from craving, fluoxetine at antidepressant doses is able to prevent relapses in weaned alcoholics. The anticompulsive therapy can positively influence the short-term outcome, while other factors are negatively associated with abstinence.