Serial 3-dimensional dental model superimposition provides a risk-free, detailed evaluation of morphological alterations on a patient’s mouth. Here, we evaluated accuracy and precision of five palatal areas, used for superimposition of maxillary 3D digital dental casts. Sixteen pre- and post-orthodontic treatment dental casts of growing patients (median time lapse: 15.1 months) were superimposed on each palatal area using the iterative closest point algorithm. Area A (medial 2/3 of the third rugae and a small area dorsal to them) was considered the gold standard, due to high anatomical stability. Areas B, C, and D added a distal extension along the midpalatal raphe, an anterior extension to the second rugae, and the remaining palatal surface, respectively. Area E was similar to A, located more posteriorly. Non parametric multivariate models showed minimal or no effect on accuracy and precision by operator, time point, or software settings. However, the choice of superimposition area resulted in statistically significant differences in accuracy and clinically significant differences in detected tooth movement (95% limits of agreement exceeding 1 mm and 3°). Superimposition on area A provided accurate, reproducible, and precise results. Outcomes were comparable for area B, but deteriorated when alternative areas were used.