Numerical simulation is increasingly being used in the field of biomechanics to predict the response of the human body to traumatic loading. Techniques such as finite element analysis and multi-body modelling are being widely applied to analyse and predict the occurrence of injury to various body parts and the results are now being used in the design of new dummies and the derivation of injury criteria.

This paper discusses the considerations that must be made when examining models of human injury, and exposes areas where knowledge of human tissue and human response is lacking to the extent that such models cannot be created without an unacceptably high degree of uncertainty in the results.

Despite these limitations, numerical simulation is a vital biomechanics research tool and examples are provided demonstrating how numerical models of different parts of the body have been effectively developed and how the results have been interpreted in the light of the assumptions on which they are based.

There is now a need to direct research effort into the development of fundamental data for improved modelling accuracy and this paper aims to inform those outside of the modelling community of the limitations of these models and the considerations that must be made when assessing their results. It also aims to suggest the directions that future research should take in order to provide the data that are essential for improving the usefulness of these models.