Electromyographic [EMG] biofeedback was compared to simple exercise therapy as to its effectiveness in improving foot-drop in 22 stroke patients. The study was designed to be a rigorous trial of biofeedback; the patients were aged and had stroke of long duration. One group of 11 patients underwent 6 wk of exercise therapy, 2 sessions/wk for 15 min/session; the 2nd group of 11 patients underwent similar therapy with EMG feedback. All therapy was conducted by a research assistant who was not a trained therapist. The groups were assessed blind before treatment, after treatment and at a 6-wk follow-up. The significantly greater improvements in the biofeedback group in terms of muscle strength at the end of treatment were maintained at follow-up. On the range of movement and gait analysis measures, both groups showed some improvement after treatment. At follow-up, this improvement had relapsed for the exercise group, while for the biofeedback group it had been maintained. Controlled trials are possible in biofeedback. Using patients as their own controls is not justified in view of the present findings and previous reports.