This paper summarizes the findings of a number of studies performed by General Motors Research Laboratories in which the double pair comparison method was used to determine the effectiveness of protection devices in preventing occupant fatalities. All the estimates use data from the Fatal Accident Reporting System, a data file maintained by the National Highway Traffic Safety Administration. This data file gives information on all fatal traffic crashes in the United States which occurred since January 1, 1975. lt is found that lap/shouldcr belts are (43±3)% effective in flreventing fatalities to car front seat occupants, lap belts are (18±9)% effective in prevcnting fatalities to car rear seat occupants, and motorcycle helmets are (27±9)% effective in preventing rider fatalities. The errors are one standard error, and effcctivenessm eans the fraction of a present population of fatally injured occupants not using the occupant protection device who would not have been killed had they been using the device, all other factors being equal.