Background: This study was aimed at comparing the accuracy of impressions of a reference typodont (RT) reproducing a totally edentulous maxilla made with three impression materials: polysulfide, polyether, and polyvinyl-siloxane. Methods: The RT was scanned using a desktop scanner, obtaining a reference scan. Ten impressions for each of the three tested materials were made using a mechanical device with a standardized and consistent modality. A laboratory scanner performed the digitization of each impression. We produced digital models by processing “in reverse” the scans of the physical impressions using a dedicated software, obtaining three groups (n = 10), respectively. The groups were titled: “polysulfide,” “polyvinyl-siloxane,” and “polyether”. The scans in .stl format were imported into Geomagic Control X and then compared to RT to evaluate the accuracy of each scan by calculating trueness and precision in µm. Recorded data were subjected to descriptive statistics. Results: Trueness (arithmetic proximity) values (95%CI) were: polysulfide = 249.9 (121.3–378.5), polyvinyl-siloxane = 216.8 (123.1–310.6), polyether = 291.1 (219.9–362.3). Precision values (95% CI) were: polysulfide = 261.9 (108.8–415), polyvinyl-siloxane = 209.4 (111.9–306.8), polyether = 283 (227.9–338.1). Statistically significant differences were not detected between the means of the experimental groups, both for trueness and precision. Conclusions: The accuracy of the scans obtained from polyvinyl-siloxane, polysulfide, and polyether impressions can be considered comparable in a fully edentulous maxilla.