This paper provides a theoretical explanation for the reported car size effect in frontal collisions. The frontal crush behaviour of the car population is examined arid shows that the specific energy absorption per unit mass propel-ties of the car population are independent of car size. Examination of single car crashes and car to car collisions in this context shows that the mean deceleration experienced by a car is inversely proportional to car length, is related to the square root of the collision closing speed arid to the inverse of the fourth I-oot of mass ratio and of crush depth. It is hypottiesised based on the Gadd severity index, for any specific car population and given degree of occupant protection within this population that Relative Injury Risk is proportional to the 2.3 power of mean deceleration. The model so derived is compared with published Relative Injury Risk data on overall risk to cars of specific size within the car population and for individual car ratings for single car non-rollover crashes and collisions between similar cars and dissimilar cars, derived from analysis of real-life accident data in Germany, Sweden, Finland and the United States. Very high correlations we obtained.