Purpose: To compare the three-dimensional (3D) accuracy of conventional impressions (CIs) with digital scans (DSs) using an intraoral scanner (IOS) with intraoral scan bodies (SBs) and varying buccolingual interimplant angulations. A secondary aim was to measure the SB machining tolerance and height with and without application of torque. Materials and Methods: Three master models (MMs) with two implants simulating an implant-supported three-unit fixed partial denture for bone-level implants were used. The implants had buccolingual interimplant angulations of 0, 10, and 20 degrees. Test models for the CI test groups were made with impression copings and polyether impressions. SBs were attached to the MMs, tightened to 15-Ncm torque, and scanned by an IOS for the DS test groups (six test groups, n = 5). A coordinate measuring machine measured linear distortions (dx, dy, dz), 3D distortions (dR), angular distortions (dϴx, dϴy), and absolute angular distortions (Absdϴx, Absdϴy) of the physical CI test models and STL files of the DS virtual models relative to the MMs. Metrology software allowed both physical and virtual measurement of geometric targets that were comparable and allowed computation of relative displacements of implant centroids and axes. Results: Mean dR ranged from 31 ± 14.2 to 45 ± 3.4 μm for DS and 18 ± 8.4 to 36 ± 6.5 μm for the CI test groups. Mean Absdϴx ranged from 0.041 ± 0.0318 to 0.794 ± 0.2739 degrees for DS and 0.073 ± 0.0618 to 0.545 ± 0.0615 degrees for the CI test groups. Mean Absdϴy ranged from 0.075 ± 0.0615 to 0.111 ± 0.0639 degrees for DS and 0.106 ± 0.0773 to 0.195 ± 0.1317 degrees for the CI test groups. Two-way analysis of variance showed that the impression technique (P = .012) and implant angulations (P = .007) had a significant effect on dR. Distortions were mostly in the negative direction for DS test groups. Perfect coaxiality of the SB with the implant was never achieved. For SB to implant machining tolerances, the mean absolute horizontal displacement ranged from 4 ± 1.2 to 7 ± 2.3 μm. The SB dz was -5 ± 3.2 μm, which increased in the negative direction to -11 ± 4.9 μm with torque application (P = .002). Conclusion: Distortions were found for both DS and CI test groups. The best accuracy was obtained with CIs for parallel implants. With angulated implants, conventional and DSs were not significantly different. Excessive torque application that causes negative dz for SB fit would position the virtual implant at a deeper location compared with reality, resulting in possible framework misfit.