Purpose. To evaluate how implant analog design and printing layer thickness affect the linear and angular accuracy of implant analogs in additively manufactured casts, comparing with conventional implant analogs in stone casts. Material and Methods. A reference cobalt chromium mandibular model with a single implant was digitized with an industrial optical scanner and scan bodies compatible with a pressure/friction fit (S) or a screw-retained (N)  implant analog for direct digital workflow. These scans were used to fabricate casts with 50 μm (S-50 and N-50) and 100 μm (S-100 and N-100) layer thickness (n=10). Ten stone casts were made after single-step closed-tray polyvinylsiloxane impressions of the model (CNV). All casts were digitized with the same metal scan body and scanner used to digitize the master model, and these scans were superimposed over the scan of the master model to measure the linear (x, y, and z-axes) and angular (XY and YZ planes) deviations (Geomagic Control X). The precision of measured deviations was defined with the average deviation values. Generalized linear model analysis was used to compare the deviations within implant analogs for direct digital workflow, while 1-way analysis of variance and Dunnett’s test were used to compare these analogs and conventional analogs (α = .05) Results. The analog design affected the linear deviations (y-axis), while the interaction between the analog design and the layer thickness, and the analog design affected the angular deviations (XY plane, P ≤ .030). S analogs had lower linear and angular deviations than N analogs, and S-50 led to lower angular deviations than N-50 (P ≤ .030). CNV led to higher linear accuracy (y-axis) than N-50, N-100, and S-100 and led to lower angular deviations than all test groups (XY plane) (P ≤ .025). Conclusions. The analogs in S-50 casts had positional trueness similar to or higher than those in other test groups and their accuracy was mostly similar to those in CNV casts. Implant analogs for direct digital workflow deviated more towards lingual and gingival, and conventional analogs deviated more towards buccal, occlusal, and distal. All analogs had a tendency to tilt towards lingual and distal.