We experienced 6 cases of proximal humeral fractures. We reviewed the facts on what fracture form the conventional classification method could support. The 5 classifications (Codman, Neer, Jakob, AO and Tamai classification) were used. In the Codman classification, we defined the type's number clockwise from upward left. Dividing the proximal end of the humerus into four segments, there were 14 patterns of the fracture. Then, the five conventional classifications were re-classified by these 14 fracture patterns. Cases 1 and 2 classified into type 3 by Codman's system and group 2 by Tamai's. There was not equivalent to the type of Neer classification. Case 1 did not apply and only case 2 fitted into Jakob's classification, because the type C2.1 was an impacted fracture. Oppositely, in the AO classification, only case 1 fitted the type C3.2 because it was a fracture with a dislocation. Cases 3 and 4 were not equivalent to Neer's, Jakob's or AO classifications. They were classified into group 3 in Tamai's system and type 15 in Codman's. Case 5 was classified into the B3.1 in the AO system and type 6 in Codman's. Case 6 corresponded only to a type 14 in Codman's classification. Because it was difficult to judge whether a fracture had occurred or not and to measure the degrees of transfer of the fracture, the reproducibility of the classification sometimes became a problem. Moreover, if no classification type was adequate for a fracture pattern, the fracture could not be consequently classified. Therefore, we were confident that the classification method should contain all fracture patterns.
