Current finite element models of head impact involve a geometrically simplified fluid-filled shell composed of homogeneous, linear and (visco) elastic materials as the primary surrogate of the human skull and brain. The numerical procedure, which solves the mechanical response to impact, requires and presumes continuity of stress and displacement between elements, a defined boundary condition simulating the neck attachment and a known forcing function.Our critical review of the models discussed, primarily, the technical aspects of the approximations made to simulate the head and the limitations of the proposed analytical tools in predicting the response of biological tissue. The following critical features were identified as major factors which compromised the accuracy and objectivity of the models: - The brain was approximated by a fluid contained in an elastic or rigid shell with no provision for relative motion between the shell and fluid. The approximation is inadequate in view of clinical and experimental evidence asserting that brain injury may be strongly influenced or caused by skull deformation and relative brain motion. - The computed shell strains were greater by as much as a factor of ten in a numerical comparison with experimental data. Moreover, the fluid pressure was dissimilar in phase and amplitude, with peak pressures representing about 25% of the experimental value. - The model objectivity was compromised by using different fluid compressibilities which varied by a factor of 200 to match experimental data. No scientific justification could be found for the wide variation in fluid property. - The first three resonant frequencies were about one-tenth of the values reported for the human head. The dynamic dissimilarity was probably due to an insufficient number of elements in the model and produced an incorrect impact response. - The acceleration input was not sufficiently representative of a typical head impact. Similarily, the boundary condition representing the neck attachment was either excluded or unrealistically simulated inspite of its strong influence on the response. - The proposed brain tolerance criteria, based on fluid pressures, indicated potential errors in space and time increments of the finite element solution.Although significant progress has been made in the development and refinement of finite element models of head impact, the available computer programs represent mechanical simulations which are inadequate to establish tolerance criteria or predict head injury. In fact, the models are only a qualitative simulation of a simplified head surrogate, and as such, constitute merely the first approximation of one aspect of a highly complex biological problem.