As part of this research effort, we developed a new methodology for projecting elderly traffic crash fatalities. This methodology separates exposure to crashes from crash risk per se, and further divides exposure into two components, the number of miles driven and the likelihood of being a driver. This component structure permits conceptually different determinants of traffic fatalities to be projected separately and has thorough motivation in behavioral theory. It also permits finer targeting of particular aspects of projections that need improvement and closer linking of projections to possible policy instruments for influencing them.