BACKGROUND: Great emphasis is placed on the skeletal medio-lateral to maintain the normal adduction of the remnant femur and to ensure coronal plane stability in an ischial containment socket. Given the invasiveness of the skeletal medio-lateral measurement, an alternative approach using prediction based on non-invasive measures would be welcomed. OBJECTIVES: Determine the accuracy with which the skeletal medio-lateral dimension could be predicted using sex, stature, anterior-posterior dimension and iliofemoral angle. STUDY DESIGN: Cross-sectional. METHOD: Anthropometric measurements on 77 persons were recorded and used as input data into a standard multiple regression. RESULTS: The regression model explained 59% of the variance in skeletal medio-lateral (r(2) = 0.59) that was statistically significant (F(4,72) = 25.37, p = 0.000). Only sex contributed significantly to the prediction of skeletal medio-lateral (β = 0.67, t = 6.15, p = 0.00). The degree of error associated with the regression model (sum of squared errors = 0.009) indicated that the actual skeletal medio-lateral could be predicted within ±18 mm in 95% of the cases. CONCLUSION: The regression model is not sufficiently accurate to predict skeletal medio-lateral for clinical purposes. Accuracy of the prediction could be improved with more accurate input data from computed tomography scans and use of other independent variables that explain the unique variance not already described by the participants' sex. Clinical relevance This pilot study demonstrates potential for the skeletal medio-lateral to be predicted using non-invasive anthropometric measurements. Given this proof of concept, future investigators should use more accurate input data from computed tomography scans and identify alternative independent variables that explain the variance in the skeletal medio-lateral not attributable to sex.