A study of 117 three-part and four-part displaced proximal humeral fractures, followed for from one to sixteen years, is presented. The ages of the patients averaged 55.3 years. Treatment began with closed reduction in seventy-seven patients, the result of which were accepted in thirty-one. Open reduction was done in forty-three patients, and prosthetic replacement in forty-three patients. Their results were rated by a numerical system. Closed reduction was found inadequate for active, healthy patients in either group. This was because of uncontrollable rotatory displacement in three-part fractures and avascular necrosis of the detached head in four-part fractures. Most of the poor results of open reduction in three-part displacements were due to errors in reduction or fixation while those in four-part displacements were due to avascular necrosis of the head. It. was concluded that the preferable method for three-part fractures was open reduction and that for four-part fractures was prosthetic replacement. Using these indications, the typical result was satisfactory but imperfect and many months were required for maximum recovery. Surgical errors and technique are discussed.