I. ABSTRACTThe performance of child restraint systems (CRS) in precise and constant laboratory testing and their effectiveness in real-life-accidents may differ substantially, mainly due to the variety of interface conditions between motor vehicles and CRS and to misuse of the CRS by the consumer. These differences may grossly impair the safety performance of CRS in a traffic crash.Most accident reports available to crash researchers lack detailed data on crash severity that can be related to injury severity. However, reports usually include full information on injuries of all occupants. The paper proposes a comparison of the injury rate of restrained children and belted adult occupants, the latter “substituting” the unknown data on crash seventy. Given a sufficient number of accident reports, the effectiveness of CRS under real-life-conditions can be assessed, providing a very useful tool, particularly for CRS manufacturers.