The effect of bone-prosthesis bonding on proximal load transfer is investigated using a coupled experimental and finite element analysis on a synthetic femur. Three-dimensional finite element models for an intact femur and a femur implanted with a cementless prosthesis were constructed from the experimental models used, and the proximal femoral strains recorded for two loading conditions approximating a one-legged stance. The approach was used to investigate a press-fitted and a fully bonded bone-prosthesis structure to identify the stem-bone behaviour for both interface conditions and their implications for proximal bone load transfer. Regression slopes close to unity indicated that the finite element predictions were an accurate estimate of the experimental measurements. Physiological surface strains were recorded only when the abductor force was included in the loading. Meanwhile, experimental measurements and numerical predictions showed that a different load transfer pattern is to be expected for normally press-fitted and glued press-fitted stems. The finite element model for the treated femur, modelling both interface conditions correlated very well with the experimental model. These finite element models subsequently modified and used to analyse the effect of different interface conditions predicted a significant increase in the load transfer to the proximal calcar bone when only proximal bonding is achieved. This study suggests that information obtained for the assessment and prediction of total hip arthroplasty longevity by numerical and experimental techniques used together and in parallel is of greater value than either technique used alone. The employment of a femur analogue as featured in this study is also shown to be a suitable alternative to cadaveric specimens in such an analysis.