The digital intraoral impression has become a central part of the CAD/CAM technique. The objective of the present study was to compare the accuracy (trueness and precision) of digital impressions of the full arch with that of conventional impressions on the in-vitro model. For this purpose, a master model was acquired with a new reference scanning process, the measuring trueness of which was +/- 4.1 microm and the precision +/- 2.5 microm. On the one hand, conventional impressions and then plaster models (n = 5) were produced from this master model, and on the other hand, digital impressions were made with the Cerec AC Bluecam and the Lava COS system (each n = 5). The plaster models were also scanned with the reference scanner. The available data records were superimposed and the differences determined. The deviation from the master model defines the trueness of the impression method. The deviations of the models among one another demonstrate the precision of the method. The trueness of the impressions was 55 +/- 21.8 microm in the conventional impression group, for digital impressions with Cerec Bluecam it was 49 +/- 14.2 microm and for digital impressions with Lava COS 40.3 +/- 14.1 microm. The precision was 61.3 +/- 17.9 microm for conventional impression with Impregum, 30.9 +/- 7.1 microm for digital impression with the Cerec Bluecam and 60.1 +/- 31.3 microm for digital impression with Lava COS. These in-vitro results show that accuracy of the digital impression is similar to that of the conventional impression. These results will have to be confirmed in further clinical studies.