Forty-five migraine patients entered a program of biofeedback-assisted relaxation training. Their mean finger temperature prior to training was found to be significantly lower than that of patients with either non-migrainous headaches, tinnitus or hypertension. Furthermore, while the other three patient groups warmed their hands several degrees during the first session, the migraine patients were unable to do so.

Of the thirty-seven patients who completed ten training sessions, twenty-seven were rated as improved. A significant inverse correlation was found between the ability to warm hands during the first session and treatment success. Although patients that improved showed increased hand temperatures across training sessions, the gain scores were not significant. Age was the second best predictor of outcome with the youngest patients showing the most improvement.