Introduction Shoulder arthroplasty is a difficult procedure which, for success, is dependant on many factors as correct retroversion of the humeral implant.

An experimental set up was therefore devised, using a model to determine the actual accuracy of the retroversion obtained under ideal in vitro conditions in two different situations, one in which the proximal humerus was intact such as that encountered in osteoarthritis or P.A.R. and the other where most usual landmarks were missing such as that seen in the four-part fracture situation.

Materials and methods 106 prostheses were inserted into 106 arms (plastic bone). 54 bones were cut at the level of the surgical collar as in complex fractures and 54 were untouched as in omarthrosis. The first series was done without any guide (52 implants). Every operator has to put 3 prostheses with 30 degree retroversion according to his particular chosen method of mark, either the bicondylar line, or according to the position of the forearm. The second one was done with a jig (Neer 3) indicating 25 degrees of retroversion (54 implants).

The degree of retroversion of prostheses put is measured according to the angle between the axis passing by the previous face of the condyles of the ulna and the axis passing by the mark taken on the prosthesis (perpendicularly in the axis of the humerus).

The humeral axis, the condylar axis, the prosthesis plane and the cutting plane were determined. 3 barycentres of humeral sections determined the humeral axis. The condylar axis is determined from the 2 barycentres of the digitalized points on the anterior articular condylar surfaces. These 2 axes determine the frontal plane on which a reference mark R(x, y, z) is attached with Z lined up with the humeral shaft and X lined up on the condyles.

Discussion and results All these results show that with or without guide, the prosthesis is not inserted in the right way. Only one third of the P.T.E. ( 33 on 106) were put with a correct angle of retroversion, or between 20 and 40 °, with a maximum of 67,7 ° and a minimum of 4,4 ° (standard deviation 12, 8 !). Shoulder prosthesis, a difficult technique, may need computer help for the positioning especially for the retroversion angle.