Objective-To assess the effect of computed tomography (CT) scan protocols (radiation amounts) and fabrication methods on biomodel accuracy and variability. Sample-Cadaveric femur of a Basset Hound. Procedures-Retroreconstructions (n = 158) were performed of 16 original scans and were visually inspected to select 17 scans to be used for biomodel fabrication. Biomodels of the 17 scans were made in triplicate by use of 3 freeform fabrication processes (stereolithography, fused deposition modeling, and 3-D printing) for 153 models. The biomodels and original bone were measured by use of a coordinate measurement machine. Results-Differences among fabrication methods accounted for 2% to 29% of the total observed variation in inaccuracy and differences among method-specific radiation configurations accounted for 4% to 44%. Biomodels underestimated bone length and width and femoral head diameter and overestimated cortical thickness. There was no evidence of a linear association between thresholding adjustments and biomodel accuracy. Higher measured radiation dose led to a decrease in absolute relative error for biomodel diameter and for 4 of 8 cortical thickness measurements. Conclusions and Clinical Relevance-The outside dimensions of biomodels have a clinically acceptable accuracy. The cortical thickness of biomodels may overestimate cortical thickness. Variability among biomodels was caused by model fabrication reproducibility and, to a lesser extent, by the radiation settings of the CT scan and differences among fabrication methods.