One hundred and sixty patients (mean age 39.8 years; 67% female) diagnosed with generalized anxiety disorder (GAD) who had completed a prospective, 8 week, double-blind comparison of lorazepam (mean daily dose 4.2 mg) and ipsapirone (mean daily dose 19.5 mg) were rapidly tapered by a substitution of half-strength medication for 3 days, then substitution of matched placebo for an additional 11 days. Patients treated with ipsapirone showed no rebound anxiety on discontinuation, nor any other significant increase in withdrawal symptomatology compared to patients who had been prospectively treated with placebo. In contrast, patients treated with lorazepam showed significant emergent anxiety and/or withdrawal-related symptomatology by almost all clinical measures employed. Overall, 25% of patients treated with lorazepam showed rebound anxiety, and 40% of them utilized reserve medication because they found drug discontinuation to be intolerable. The clinical implications for discontinuation of benzodiazepines after short-term therapy are discussed.