Aim: Despite the expanding implementation of intraoral scanning (IOS) devices, indirect digitization of conventional impressions or casts still represents the primary access to CAD/CAM. The aim of this study was to evaluate the accuracy of data acquired from impression scans and cast scans with respect to impression material and type of cast used. Materials and methods: A standardized titanium model for a four-unit fixed dental prosthesis (FDP) served as a testing model. Industrial computed tomography (CT) was applied, generating a reference data set. Four different impression materials were utilized (n = 12 per material): 1) Impregum Penta (polyether/group PE); 2) Imprint 4 Penta Super Quick Heavy + Super Quick Light (polyvinyl siloxane (PVS)/group PVS-I); 3) Dimension Penta H Quick + L (PVS/group PVS-D); and 4) Imprint 4 Preliminary Penta Super Quick (PVS/group PVS-P). Data were obtained from three different model situations, ie, impressions (group IMP), unsectioned plaster casts (group UNSEC), and sectioned casts (group SEC). The surfaces were digitized three times each using a laboratory scanner. The resulting test data were superposed with the reference data using a best-fit algorithm to evaluate accuracy. Statistical analysis was conducted using the Kolmogorov-Smirnov, Kruskal-Wallis, and Mann-Whitney tests (level of significance: P 0.050). Results: Imprint 4 Penta presented the highest overall accuracy, while Imprint 4 Preliminary Penta Super Quick displayed the poorest results. Regarding the model situation (impression scan vs cast scan), impression scans from Impregum Penta and Imprint 4 Penta showed superior results. Conclusion: Impression scans in combination with high-precision impression material results in the most accurate data. Keywords: accuracy, CAD/CAM, cast scan, digital impression, impression material, impression scan, STL data