Background: Abrupt cessation of alcohol intake causes habituated drinkers to experience symptoms of alcohol withdrawal syndrome (AWS). Objective: To determine the effect of the gamma-aminobutyric acid (GABA)-B agonist baclofen on the course of acute symptomatic AWS. Design: Prospective, randomized, double-blind, placebo-controlled clinical study. SETTING:: Two tertiary-care hospitals in Duluth, Minnesota. Patients: Inpatient adults admitted for any reason (including AWS) judged to be at high risk for AWS. Intervention: Inpatients who developed symptoms of AWS received symptom-triggered benzodiazepine treatment using lorazepam by standard protocol, and were randomized to receive baclofen 10 mg or placebo, 3 times per day, orally. MEASUREMENTS:: AWS severity was assessed using the Clinical Institute Withdrawal Assessment of Alcohol Scale, Revised (CIWA-Ar); lorazepam dose was monitored. RESULTS:: Seventy-nine subjects were enrolled. The 44 subjects who developed symptoms of AWS were randomized to baclofen or placebo. Thirty-one subjects (18 baclofen, 13 placebo) completed 72 hours of assessments, either entirely as inpatients or with outpatient follow-up. The need for high doses of benzodiazepines (20 mg or more of lorazepam over 72 hours) to control AWS was less likely in the baclofen treatment group (1 of 18) than in the placebo-treated group (7 of 13) (P = 0.004). Conclusions: We found that the use of baclofen was associated with a significant reduction in the use of high doses of benzodiazepine (lorazepam) in the management of symptomatic AWS. The use of low-dose baclofen in the management of AWS deserves further study, as reduced dependence on high-dose benzodiazepines in AWS management could improve patient safety.