Accurate measurement of the pelvis is critical for well-fitting and comfortable ischial containment sockets. The "Skeletal Medial-Lateral (ML)" is intrusive and unreliable to measure in vivo. This study aimed to determine how accurately the Skeletal ML could be predicted and to identify which measurements were significant predictors. Computed tomography scans were randomly sampled from a cadaveric database (n = 200). Inclusion criteria were age > 20 yr; lower-limb alignment that replicated the anatomical position; and no evidence of osteological trauma, implants, or bony growths. Multivariate linear regression models were developed to predict the Skeletal ML based on a suite of independent variables, including sex, body mass, and distance between pelvic landmarks. The regression model explained 76% of the variance in the Skeletal ML (p < 0.001). Variables that contributed significantly to the prediction of the Skeletal ML (p < 0.05) included body mass, sex, inter-greater trochanter distance, pelvic depth, and age. Significant predictors of the Skeletal ML dimension characterize variation in subcutaneous adipose tissue thickness and pelvic morphology. The Skeletal ML could be predicted with relatively small errors (standard error of the estimate = 7 mm) that could be easily and reliably adjusted during socket fitting. Further research is needed to test the predictive tool in a real-world setting.