For some time the need for more consistent, more realistic and more complete reporting of injury severity in the national road accident statistics has been recognised. The study reported, which was based on over 3600 casualties seen at one large accident hospital in Berkshire, examines the use of length of stay as a simple objective criterion of severity for use in this context, by non-medical personnel. Length of stay has been related to clinical severity as defined by the abbreviated injury scale (AIS). A simple three-category classification is suggested: not detained, detained for less than 3 nights, and detained for 3 nights or more. An examination of the under-reporting of casualties to the police is also made. In the area covered nearly 30 per cent of casualties were not reported to the police, the degree of under-reporting being greater for pedestrians and two-wheeler riders than for vehicle occupants. /Author/TRRL/