Seventy-eight patients who had a complete brachial plexus palsy caused by a stretching injury were operated on by the same surgeon between January 1980 and December 1991. The aim of the operative intervention was to obtain the best functional result, including at the level of the hand, that was possible in view of the initial lesions and the intraoperative findings. Therefore, the treatment strategy included not only nerve repair with grafting (124 grafts) or nerve transfer (twenty-seven transfers) but also palliative procedures, the latter of which sometimes were performed several years later. Sixty-three patients were evaluated by an independent observer at least three years postoperatively. The results associated with each type of lesion and each type of nerve repair were assessed according to the function of the muscles that were innervated by the recipient nerve. Six patients had a neurolysis only. The remaining fifty-seven patients had grafts or nerve transfers to repair the biceps. Thirty-six of the fifty-seven received a rating of 3+ or more (meaning that the patient was able to flex the elbow repeatedly); the remaining twenty-one received a rating of 3 or less (meaning that the patient was able to flex the elbow only once or not at all), which we considered unsatisfactory. The function of the triceps recovered after eleven of thirty-one procedures that were performed to restore that nerve; that of the extensor carpi radialis, after five of thirty-one procedures; that of the flexor carpi radialis, after six of thirty-one procedures; and that of the flexor digitorum, after four of thirty-one procedures. A statistical analysis revealed that an operative delay of less than six months was a significant factor with respect to recovery of the function of the biceps (p = 0.003). The thirty-nine grafts that were sutured onto the lateral or posterior cord produced better results than did the thirty-six that were sutured onto the distal branches (the musculocutaneous and radial nerves); however, with the numbers available, this difference was not found to be significant (p = 0.08). Eleven patients had a successful result (a rating of 3+ or more) and eight, a fair or poor result, with respect to recovery of biceps function after transfer of the spinal accessory nerve to the musculocutaneous nerve. Overall, twenty-nine patients had relief of pain postoperatively. Sixteen patients had grade-3 pain preoperatively compared with only three after the operation. According to a self-rating scale, twenty-five patients were satisfied with the overall result, sixteen were fairly satisfied, and twenty-two were dissatisfied.