The many measuring methods of humeral head retrotorsion angle using CT were reported. The purpose of this study was to compare the methods which were reported in the past with our original method. Pneumo-arthro CTs of two hundred of one hundred patients with unilateral posttraumatic recurrent dislocation of the shoulder were examined. Eighty-seven patients were male and thirteen were female. We measured the retrotorsion angle using CTs at the levels of the humeral head and the distal humerus. We adopted three kinds of lines at the level of the proximal humerus: the S line (Simeonides' Method), the L line (Laumann's Method) and the A line our original method. We also adopted two lines at the level of the distal humerus: the T line (Tosu's Method) and the R line (Randelli's Method). We measured the humeral retrotorsion angle using a combination of the proximal three lines and the distal two lines. We also examined the reproducibility of three proximal methods by multiple measurements and the measurement by multiple testers. The retroversion angle using a combinationof the S line and the T line was 29.6±12.1°(-5.5 to 59.4°), and using acombination of the S line and the R line was 27.0±129°(-72 to 63.3°). The results using a combination of the L line and T or R line were 29.3±22.3°(-292 to 84.3°) and 27.0±22.1°(-24.5 to 76.2°). The results using a combination of the A line and the T or R line were 30.4±10.9°(6.2 to 57.5°) and 28.3±11.2°(2.1 to 57.0°). The different of the mean value of the retrotorsion angle were not significant by any kinds of measuring method atthe level of the humeral head. However, the standard deviations of results using the L line method were extremely bigger than those using the S and A lines. The L line method showed an extremely high and low angle compared with the other two methods, and seemed to be less reproductive. Our original method using the A line was the most easily reproducible. The mean value of the retrotorsion angles were approximate whichever method was used at the level of humeral head, but the standard deviation of the result by using the L line method was extremely bigger than those using the S and A lines. Our method using the A line was most easily reproducible compared with the methods using the S and L lines.
