The problem of establishing accident involvement rates for road vehicles is a complex one. Difficulties derive from the large number of factors-other than those connected with the vehicle only-in the man-vehicle-road environment complex. Proving ground testing of vehicles may lead to results of limited applicability. Long-term performance on highways is the ultimate criterion of safety.The results of many studies indicate that accidents attributable to direct vehicle defects amount to only 8% of the total. Study of these data alone is not sufficient to establish the safety performance of a particular vehicle.To establish vehicle accident involvement rates, both for particular vehicles and for systems of vehicles, the authors present a model that has been simplified to include traffic density, vehicle deployment, and driver age-three important factors for which epidemiological data are available or may be obtained. In addition, using a novel graphic technique, comparisons of vehicles for variations in accident rates and exposure (mileage) are performed with ease. The exposure-accident graphs of this technique permit meaningful analysis of accident rate trends.Examples are presented for classes of vehicles operating in the United States and Israel, and a “case study” is made for evaluating the accident involvement rate of a typical 2-door, medium-priced passenger car.