The aim of this study was to determine the validity with which the finite element method could model synthetic bone and thereby determine the appropriateness of such femur analogues for application in pre-clinical tests. The performance of these synthetic femora was compared with cadaveric bone when employing the same geometric and material definition protocols. A four-point bend loading configuration was selected for this analysis. Four synthetic femurs and an embalmed cadaveric bone were tested experimentally to determine the structural bending stiffness (k) for the diaphysis of these bones. A finite element (FE) model was generated and an analysis performed for each bone type to estimate the Young's modulus (E) required to obtain a model stiffness equivalent to that obtained experimentally. The estimated material elastic modulus in the FE model for the synthetic femur was found to be very similar to available data for this bone analogue. The estimated cadaveric bone modulus however was found to differ significantly from documented values for cortical bone. A theoretical analysis demonstrated the great sensitivity of the estimated modulus value to the accuracy of the geometric definition. The very low variability found in the experimental test on the synthetic bones together with their more regular geometry and the possibility of achieving greater accuracy in geometric definition was shown to enable the production of a valid FE model of this bone for an isotropic homogeneous material description. Conversely, the greater irregularity of geometry, together with the less obvious differentiation between the cortical and cancellous bone in the cadaveric specimen makes accurate geometric description of this bone very difficult. This fact, together with the uncertainty concerning the quality of the cadaveric bone and its viscoelastic response during mechanical testing, makes reproduction of its behaviour in a FE model a much more demanding task. It is suggested that this greater capability of reproducing the experimental behaviour of the synthetic bone makes them a very useful model for both experimental and numerical studies which involve in-vitro pre-clinical testing of implant design and stem-bone behaviour.