In a double-blind, randomized, crossover study, 58 patients with postherpetic neuralgia received 6-week courses of amitriptyline, 12.5 to 150 mg/d; lorazepam, 0.5 to 6 mg/d; or lactose placebo. Doses were titrated to the maximum level tolerated. Patients rated pain in a diary, using lists of verbal descriptors. Forty-seven percent of patients reported moderate or greater relief with amitriptyline, 16% with placebo, and 15% with lorazepam. Mean amitriptyline dose was 65 mg/d. Greater relief was associated with higher amitriptyline doses, up to the maximum dose of 150 mg/d, and with higher serum tricyclic levels. Lorazepam did not relieve pain and was associated with severe depressive reactions in four patients.