The response to diazepam and propranolol hydrochloride was compared in 21 patients who (with one exception) met DSM-III criteria for panic disorder and agoraphobia. Each drug was administered for two weeks in double-blind fashion according to a crossover design. The response to diazepam was significantly superior on all measures. By observer rating, 18 patients showed at least moderate improvement with diazepam compared with seven receiving propranolol. Panic attacks and phobic symptoms responded to diazepam, but not to propranolol. The results suggest that benzodiazepines constitute effective short-term treatment for these newly defined disorders.