Throughout literature a number of glenoid classification systems have been described but lack clear correlation with the fracture patterns found in clinical cases. This study aimed to evaluate the intra- and interobserver agreement for fracture classifications of the glenoid, using either plain radiographs, computed tomography (CT) scans. The study was retrospective, using images with a variety of fracture types. Six observers classified the fracture patterns. Classifications of Ideberg, OTAAO, Mayo and Euler/Ruedi were used. Agreement was determined using kappa coefficients. Currently used glenoid fracture classification systems have a fair to moderate intraobserver reliability. Combining plain radiographs and CT scans led to a better observer agreement. For interoberserver reliability, the system of Euler scored slightly better than other systems. Although Ideberg's classification is the most widely used system, this study does not support superiority of it. Based on this study there is need for a more reliable glenoid classification system.