Objective: The aim of this study was to test the hypothesis that, because of abnormalities in serotonergic neurotransmission that may underlie craving and impulsive behavior, fluoxetine treatment differentially affects drinking among type B alcoholics, who are characterized by high levels of both premorbid vulnerability and alcohol-related problems. Methods: Using a k-means clustering procedure, alcohol-dependent subjects from a placebo-controlled trial of fluoxetine were grouped into low-risk/severity (type A n= 60) and highrisk/severity (type B: n= 35) groups. Multivariate analysis of covariance (with pretreatment measures as covariates) evaluated the effects of Alcoholic Subtype, Medication Group, Treatment Completion, and their interactions on measures of drinking, both during the 12-week treatment period and a 6-month follow-up period. Results: Although there were no main effects of Alcoholic Subtype or Medication Group, subjects who completed the treatment trial showed significantly better drinking-related outcomes. There was also an interaction of Alcoholic Subtype by Medication Group during treatment. Among type B subjects, fluoxetine treatment resulted in poorer drinking-related outcomes than placebo treatment. Among type A subjects, there was no effect of Medication Group. This interactive effect did not persist during the 6-month follow-up period. Conclusions: Alcoholic subtypes identified by cluster analysis seem to be differentially responsive to the effects of fluoxetine treatment on drinking-related outcomes. Serotonergic abnormalities previously identified among a subgroup of alcoholics who are also characterized by impulsivity and severity of alcohol dependence may help to explain the differential medication effect. Based on these findings, it is recommended that, in the absence of a comorbid mood or anxiety disorder, fluoxetine not be used to maintain abstinence or reduce drinking in high-risk/severity alcoholics.