Aims: The aim of this study was to compare two existing methods and one novel method for measuring the distortion of three-dimensional (3D) models created with complete-arch digital impressions, and to assess the accuracy of different scan patterns using these methods. Materials and methods: Maxillary and mandibular models were imaged with the PlanScan intraoral scanner using four different scan patterns. Accuracy and distortion were assessed by comparing the master scans with the intraoral scans using the following three methods: 1) Mean surface deviation was measured after complete arch superimposition; 2) 28 points were selected identically on the experimental and on the master reference models, and the deviation between identical points was assessed after superimposition over the complete arch; 3) In the case of the novel technique, the superimposition was made only at the scanning origin, and after that the 28 points were compared. Results: Significant differences were found between the three different methods, regardless of the arch and pattern. The overall mean deviation between identical points when the models were aligned at the scanning origin was the highest, and the mean deviation between the non-identical values was the lowest. The novel method revealed local tooth-wise differences between the scan patterns as well as a pattern of amplified model error extending away from the scan origin. Conclusions: The novel method better detects the cumulative deviation of stitching errors in complete arch intraoral scans and is suitable to investigate the effect of scanning pattern in a very sensitive manner. Keywords: digital impression, intraoral scan, accuracy, scan pattern, deviation, stitching, registration, distortion