The specificity of EMG biofeedback (myofeedback) was examined by comparing the effects of actual myofeedback, simulated myofeedback and no myofeedback in the treatment of foot drop and shoulder subluxation. A group of twenty-four hemiplegic subjects, who were simultaneously undergoing traditional physical therapy, were randomly assigned to either Group I (Actual Myofeedback Group; AMG) or Group II (Simulated Myofeedback Group; SMG). Those in Group I received ten sessions of actual myofeedback over a two week period. Those in Group II received a similar course of treatment but without their knowledge were given a simulated myofeedback initiated by the therapist corresponding to the patient's apparent efforts rather than their actual performance or muscle activity. For those subjects in whom both deltoid and anterior tibial muscles met the study criteria, only one muscle was treated as a part of Group I or Group II. The other muscle was assigned to Group III (Control Group; CG) and evaluated before and after a two week period during which it underwent no myofeedback therapy. Upon statistical evaluation, AMG, SMG and CG were all found to have statistically significant increases in muscle function in terms of both mean change in muscle activity measured in microvolts and in mean change in active range of motion measured in degrees. While no significant difference was found between AMG and SMG in either of these parameters, both AMG and SMG were found to have significantly greater increases than CG. It was concluded that although myofeedback appears to be an effective addendum to traditional physical therapy for hemiplegic patients, its success is due to as yet undefined elements of the treatment design.