Background: The aim of this study was to evaluate linear measurements in digital models, scanned at different angulations, in order to compare four methods of superimposing digital models. Methods: Dental models obtained from the database of the School of Health and Life Sciences of the Pontifical Catholic University of Rio Grande do Sul (PUCRS) were scanned three-dimensionally at different angulations (0˚, 5˚, 10˚ and 15˚ in relation to the occlusal plane) and divided in four groups. The sample size was calculated and 43 dental models were used in this study. Linear measurements were obtained to assess the central and lateral portions of the 3D models, so that any possible distortion would be detected. Additionally, the digital models were superimposed using four approaches based on anatomical references: Methods 1 (three central landmarks), 2 (three lateral landmarks), 3 (surface + single landmark) and 4 (surface + three landmarks). Intraclass correlation coefficient was applied to evaluate the intra-observer reliability. Kruskal-Wallis was calculated to compare the groups according to the angulation during scanning, and Kappa to evaluate the agreement between superimposition methods. Results: No differences were found in models scanned at different angulations. Better superimposition correlation was observed between methods 3 and 4 those used techniques based on palatal surface associated to anatomical landmarks to superimpose. Conclusions: Linear measurements and superimpositions were not affected by the different angulations of the models during the scanning process. The most reliable superimposition methods were the ones using surface and landmarks.