Purpose: To evaluate the in vitro accuracy of impressions obtained with two silicone and corresponding stone models using two laboratory scanners. Materials and Methods: A master model with synthetic resin teeth with two single-unit crown preparations was created and scanned using a 12-megapixel scanner. Five conventional impressions of the physical model were prepared with different silicone impression systems (Zhermack and Coltene) using the double-mix technique and poured with gypsum. The impressions and stone models obtained were scanned using two extraoral scanning systems (Identica T500, Medit; S600 ARTI, Zirkonzahn). All best-fit superimpositions of the teeth areas were conducted between the master model and the scans of the impressions and models obtained with the two scanners. A P < .05 level was considered significant. Results: The Identica T500 Medit scanner  showed an accuracy of 102.34 (89.67, 115.01) μm for Coltene silicone and 79.51 (67.82, 91.21) μm for Zhermack silicone, while the S600 ARTI Zirkonzhan scanner presented 110.79 (98.24, 123.33) μm and 91.91 (81.29, 102.54) μm, respectively, with significant differences between scanners for Zhermack silicone (P = .008) and for the corresponding stone models (P = .002). Zhermack silicone presented overall discrepancies lower than Coltene silicone, with statistically significant differences in both scanners analyzed (P < .001; P = .017). However, the discrepancies found were within clinically acceptable values. With the Zirkonzahn scanner, discrepancies found in the Zhermack impressions were lower than in the corresponding stone models (P < .001). Conclusions: The direct digitization of silicone impressions with laboratory scanners presented comparable results to conventional techniques with stone models.