OBJECTIVE: This study was designed to examine the possibility that patients with panic disorder are especially vulnerable to alprazolam withdrawal, as well as to evaluate the efficacy of carbamazepine as adjunctive treatment during alprazolam discontinuation. METHOD: After a 2-month open trial of alprazolam, 36 patients with panic disorder and 35 with generalized anxiety disorder entered a controlled discontinuation phase. Carbamazepine or placebo was added in a randomized, double-blind fashion, followed after 1 week by single-blind dose reduction of alprazolam, approximately 25% every third day. RESULTS: Fifty-two percent of the patients were able to discontinue alprazolam, but only 37% completed the study, maintaining alprazolam- free status for 4 weeks. Survival analysis revealed that among the patients receiving placebo as adjunctive therapy, panic disorder patients had a significantly greater dropout rate than patients with generalized anxiety disorder. While carbamazepine exerted no beneficial effect for patients with generalized anxiety disorder, it appeared to improve outcome in the panic disorder group. CONCLUSIONS: Results of this study indicate that panic disorder patients are more vulnerable to alprazolam withdrawal than patients with generalized anxiety disorder, and this may reflect a general diathesis of panic disorder patients to encounter more difficulty during drug withdrawal. Carbamazepine may selectively assist with alprazolam withdrawal in panic disorder patients. However, these data do not support its widespread use during benzodiazepine discontinuation, especially given its potential toxicity. In light of a proposed greater vulnerability of panic disorder patients to withdrawal, future studies need to account for differential effects related to specific diagnosis.