Many elastomeric impressions sent to commercial laboratory dental technicians may include marginal defects. To fabricate accurate restorations, digital technology may be used to merge digital files of defective impressions into a single standard tessellation language (STL) file free of errors. This would save clinicians and patients time and may improve clinical care. The purpose of this study was to compare the accuracy of digital master casts reconstructed from merged STL files of defective impressions with the file of the original defect-free preparations. Ivorine teeth on a dentoform were prepared to receive a posterior fixed dental prosthesis (FDP) with complete coverage preparations. An impression was made in a stock tray using polyvinyl siloxane (PVS) impression material and an extraoral scanner (E3, 3Shape, Denmark) was used to digitize the impression; this was the reference cast. Wax was used to create defects on the buccal and lingual margins of the preparations. Fifteen PVS impressions were made of the FDP preparations with defects in the mesial and distal margins; another set of 15 PVS impressions was made of FDP preparations with defects in the buccal and palatal margins for a total of 30 impressions. All impressions were digitized using the same extraoral scanner (E3, 3Shape, Denmark). Corresponding STL files were paired and merged, and a master cast was created by eliminating the defects using the scanned data. This master cast was compared to the reference cast using reverse engineering software (Geomagic, Morrisville, NC, USA). The results were expressed as average errors and standard deviations in the master casts relative to the reference cast. To account for the presence of positive and negative values in the data set, in terms of errors, the root mean square (RMS) value was calculated for each sample. The mean average error in the sample was -0.4 μm. The average upper limit of 95% confidence interval was +36.5 μm, while the average lower limit of 95% confidence interval was -37.3 μm. The mean RMS of the errors found was 18.9 μm. The results of this study indicated that merging digitized definitive impressions to correct marginal defects resulted in master casts with a high level of accuracy relative to the reference cast.