Composite synthetic models of the human femur have recently become commercially available as substitutes for cadaveric specimens. Their quick diffusion was justified by the advantages they offer as a substitute for real femurs. The present investigation concentrated on an extensive experimental validation of the mechanical behaviour of the whole bone composite model, compared to human fresh-frozen and dried-rehydrated specimens for different loading conditions. First, the viscoelastic behaviour of the models was investigated under simulated single leg stance loading, showing that the little time dependent phenomena observed tend to extinguish within a few minutes of the load application. The behaviour under axial loading was then studied by comparing the vertical displacement of the head as well as the axial strains, by application of a parametric descriptive model of the strain distribution. Finally, a four point bending test and a torsional test were performed to characterize the whole bone stiffness of the femur. In all these tests, the composite femurs were shown to fall well within the range for cadaveric specimens, with no significant differences being detected between the synthetic femurs and the two groups of cadaveric femurs. Moreover, the interfemur variability for the composite femurs was 20-200 times lower than that for the cadaveric specimens, thus allowing smaller differences to be characterized as significant using the same simple size, if the composite femurs are employed.