A modified Bristow-Helfet-May procedure was performed for recurrent dislocation or subluxation of the glenohumeral joint in 207 patients (212 shoulders), whose average age at the time of surgery was 20.3 years (range, fourteen to forty-seven years). The procedure was modified by directing the coracoid segment and conjoined tendon over the superior border rather than through the substance of the subscapularis tendon and muscle. The indications were either documented recurrent anterior dislocation of the glenohumeral joint or subluxation with instability as demonstrated on examination with the patient under anesthesia. The procedure was evaluated on the basis of the rate of recurrence of dislocation and subluxation, postoperative complications, the patients' subjective evaluation, and the effect of the procedure on the motion of the glenohumeral joint and the strength of the muscles of the shoulder as related to overhead throwing. Eight (3.8 per cent) of the shoulders redislocated and ten (4.7 per cent) had one or more subjective episodes of subluxation after the procedure. Complications included postoperative infection in two patients and problems with the screw that required its removal in ten. One hundred and thirty-one (62 per cent) of the patients responded to a questionnaire regarding their subjective evaluation of the results of surgery. Eleven (8 per cent) were unable to perform daily activities that involved overhead work, and forty-five (34 per cent) stated that they still had some degree of discomfort or pain in the shoulder. One hundred and twenty-six patients (96.2 per cent) stated that they were happy with the results of the surgery and would have the procedure again. Thirty patients had Cybex testing of the muscles of the shoulder. Only three (16 per cent) of the nineteen athletes whose dominant arm had been operated on returned to their pre-injury level of throwing. Data obtained with regard to changes in the range of motion and strength of the glenohumeral joint indicate that this loss of throwing ability was not due solely to a loss of glenohumeral motion. It appeared to be also related to a concomitant loss of strength at the extreme of external rotation of the humerus and the initiation of internal rotation of the humerus.