Book excerpt: State-level studies of direct motor vehicle accident costs are based on data which are out of date. If these data are used, the Texas Transportation Institute's 1982 update is most accurate and more current than the State-level cost data presented in the AASHTO "Red Book" (1977) and the Transportation Research Board's Methods for Evaluating Highway Safety Improvements (1975). Good national direct cost estimates (and estimates of the cost of police, coroners, etc.) appear in: The Incidence and Economic Costs of Major Health Impairments by Hartunian, Smart, and Thompson (1981) and The Economic Cost to Society of Motor Vehicle Accidents by NHTSA (1983). Good estimates for human capital costs appear in NHTSA (1983) and are preferred over the National Safety Council's 1981 estimates. No estimates exist of psychosocial costs. Estimates of willingness to pay for life and safety would be theoretically superior to human capital costs for use in benefit-cost analyses. Empirical studies of willingness to pay offer widely divergent value-of-life estimates, and most are based on questionable data, assumptions, or estimation procedures. A survey is needed to determine willingness to pay in a highway safety context. Failing that, use of the willingness-to-pay/human-capital approach presented by Landefeld and Seskin in the American Journal of Public Health in 1982 is recommended. It yields willingness-to-pay values approximately 2.128 times as large as human capital costs. A 4-percent discount rate is recommended