Reports of in-depth crash studies from Australia, UK, Europe and the USA were reviewed. Successful studies were found to have defined and explicit objectives, a sound statistical design, appropriate analysis, and a substantial report which was related to the objectives of the study. On-scene studies gather transient information, and allow development of insight into the crash process; follow-up studies are more convenient, but at the cost of 1st information. The resources required for each are similar. Apart from a small rural study near Adelaide, there has been no field study of crash performance of vehicles in Australia since 1977. There is a need for a system for continual monitoring of the effects of changes in vehicles and the environment on injuries and the crash process. This should be done through on-scene studies of rural and urban crashes, and through a system of augmented police reports.