In a randomized, double-blind crossover study, 29 patients with painful diabetic neuropathy received 6 weeks of amitriptyline and 6 weeks of an "active" placebo that mimicked amitriptyline side effects. Amitriptyline was superior to placebo in relieving pain in weeks 3 through 6. Both steady, burning pain and lancinating pains were relieved. Patients able to tolerate higher amitriptyline doses reported greater relief, through the maximum dose of 150 mg nightly. Amitriptyline analgesia was similar in depressed and nondepressed subgroups and was not associated with mood improvement. We conclude that amitriptyline relieves pain in diabetic neuropathy; this effect is independent of mood elevation.