Sixty-four headache sufferers were allocated randomly to cognitive-behavioral therapy (CBT), temporal pulse amplitude (TPA) biofeedback training, or waiting-list control. Fifty-one participants (14M/37F) completed the study, 30 with migraine and 21 with tension-type headache. Treatment consisted of 8, 1-hour sessions. CBT was highly effective, with an average reduction in headaches from pre- to posttreatment of 68%, compared with 56% for biofeedback, and 20% for the control condition. Headaches continued to decrease to 12 month follow-up for CBT. Improvement with CBT was associated with baseline coping skills, social support, and physiological measures at rest and in response to stress, particularly TPA. Changes on some of these measures were correlated with changes in headaches. No significant predictors of response to biofeedback emerged.