A novel methodology is presented to estimate the safety benefits of intelligent vehicle safety systems in terms of reductions in the number of collisions and the number and severity of crash-related injuries. In addition, mathematical models and statistics are provided to support the estimation of the crash injury reduction factor in rear-end, lane change, and single vehicle roadway departure collisions. Simple models based on Newtonian mechanics are proposed to derive Δv, the change in speed that a vehicle undergoes as a consequence of crashing. Statistics on the distribution of vehicle types and weights in the United States are supplied, which are needed for Δv estimation. Moreover, mathematical equations are derived to estimate the average harm per collision. Finally, statistics on the average harm per occupant are obtained from the 1994 and 1995 Crashworthiness Data System crash databases.