Purpose The main purpose of this study is to establish which of two methods is more reliable in glenoid assessment for instability in pre-operative planning. Accordingly, we have studied the intra-and inter-observer reliability of glenoid parameters with the use of two-dimensional (2D) and three-dimensional (3D) reconstructed computed tomography (CT) images. Methods One hundred glenoids were measured with the use of 2D-CT and 3D-CT (in 3D orientation) by two independent observers (one experienced and one inexperienced). Measurements were repeated after one week for 30 randomly selected glenoids. Results The intra-class correlation coefficient (ICC) for interobserver reliability was significantly greater for 3D-CT (0.811 to 0.915) than for 2D-CT (0.523 to 0.925). All intra-observer reliability values for 3D-CT were near perfect (0.835 to 0.997), while those for 2D-CT were less reliable (0.704 to 0.960). A dependent t-test showed that, for both observers, almost all glenoid parameters (except R and d) differed significantly (p < 0.05) between 2D and 3D measurement methods. Conclusions Therefore, it can be concluded that 3D glenoid reconstructions are more reliable for glenoid bone loss assessment than 2D-CT. The results suggest that quantifying a glenoid defect with the use of 2D image only-even if performed by an experienced orthopaedic surgeon-is prone to errors. Differences in measurements between and within observers can be explained by plane setting and identifying glenoid rim in 2D -C T. Accordingly, we recommend that glenoid measurements should be performed in 3D orientation using 3D reconstruction obtained from CT images for pre-operative assessments, which are crucial for surgical planning.