Introduction: Impression making and fabrication of plaster models serve as a basis for documentation, diagnosis and treatment planning. Model analysis is used routinely in orthodontics and is a key factor for treatment planning and review of orthodontic progress. Recently newer technologies in imaging and digitalisation have made it possible to obtain 3-Dimensional (3D) virtual models. Virtual models score over plaster models with respect to accessibility, storage and long life. However, there are concerns over its accuracy and reproducibility. Aim: To evaluate the accuracy of three dimensional digital models obtained from Cone Beam Computed Tomography (CBCT). Materials and Methods: This was an ex-vivo study. Thirty dental models of patients were selected and scanning of these models were done using CBCT. Means and standard deviations of the space analysis measurements were calculated by using both plaster and digital models. Statistical analyses were used to investigate the accuracy and repeatability of the methods studied using Statistical Package of Social Sciences (SPSS) 20.0. Paired t-tests for both mean and standard errors were calculated. R esults: The results for all 30 maxillary and mandibular models were similar. There was a statistically significant difference in calculating sum of ten teeth for both maxillary and mandibular teeth between the plaster and dental digital models and in mesio distal measurements between digital and plaster dental models. C onclusion: Digital models provide a more accurate measurement of mesio distal width of tooth with a statistically significant difference of (p-value=0.002).