Improved treatment of alcohol dependence is a high priority, including defining subtypes that might respond differently. We evaluated a medication combination of intravenous flumazenil (FMZ) and oral gabapentin (GBP) in alcoholics who did and did not exhibit pretreatment alcohol withdrawal (AW) symptoms. Sixty alcohol-dependent individuals (44 with low AW and 16 with high AW) were randomized to receive FMZ (2 mg of incremental bolus for 20 minutes for 2 consecutive days) and GBP (up to 1200 mg nightly for 39 days) or their inactive placebos. Alcohol withdrawal was measured for the first 2 days, and drinking, sleep parameters, and adverse events were monitored during weekly evaluations, along with behavioral counseling sessions. Percent days abstinent (PDA) during treatment and time to first heavy drinking (TFHD) day were primary outcome variables. There was an interaction between the pretreatment AW status and the medication group on PDA (P = 0.0006) and TFHD (P = 0.06). Those in the high AW group had more PDA and more TFHD if treated with active medications, whereas those in the low AW group had more PDA and more TFHD if treated with placebo. This interaction remained for those totally abstinent (P = 0.03) and was confirmed by percent carbohydrate-deficient transferrin values. In addition, the pattern of response remained up to 8 weeks after treatment. In addition, in those with high AW, greater improvement in AW symptoms was observed in the active medication group compared with the placebo group. These results suggest a differential response to FMZ/GBP treatment, depending on pretreatment AW status that should be taken into account during future treatment trials.