In a placebo-controlled, multicenter study, 180 male and female outpatients, ages 18-65, with DSM-III-R generalized anxiety disorder, were treated with abecarnil (a partial benzodiazepine agonist), alprazolam, or placebo for 4 weeks. This was followed by a rapid (1-week) taper, during which patients were assessed for any taper-related symptoms. All patients were identified via a structured clinical interview for DSM-III-R and randomly assigned to one of the three treatment groups. More than 70% of each treatment group completed the study. In the acute-treatment phase, both abecarnil and alprazolam showed evidence for efficacy that was significantly better than that of placebo. Both active agents were tolerated well. After the swift taper, a significantly greater number of taper-related symptoms occurred in the alprazolam-treated group than in the abecarnil-treated group, which was not different than in the placebo-treated group. Additionally, less residual improvement followed the taper in the alprazolam-treated and the placebo-treated groups. These data indicate that the partial benzodiazepine agonist abecarnil may be useful as a safe, effective, short-term treatment for anxiety. Theoretical and practical implications of these findings are discussed.