The relative risks of fatal crash involvement at various blood alcohol concentrations (BACs) were examined using data on fatal driver injuries from the Fatal Accident Reporting System in conjunction with driver exposure data from the second national road-side breath-testing survey. Based on driver fatalities in single-vehicle crashes, it was estimated that each 0.02 percentage increase in the BAC of a driver with non-zero BAC nearly doubles the risk of being in a fatal crash. Crash risk was found to increase with increasing BAC among all of the six age and sex groups studied. At BACs in the 0.05-0.09 percent range, the likelihood of a crash was at least nine times greater than at zero BAC for all age groups. Younger drivers with BACs in the 0.05-0.09 range had higher relative risks than older drivers, and females had higher relative risks than males. At very high BACs (at or above 0.15 percent), the risk of crashing was 300 to 600 times the risk at zero or near-zero BACs. These relative risk estimates are considerably higher than estimated in other studies, but other studies have based their estimates on all crashes rather than single-vehicle crashes only. In this study, relative risks were also lower when based on driver fatalities in all crashes. However, when plausible assumptions were made about the BAC distributions of other participants in multiple-vehicle crashes (whose actual BAC is often unknown), the relative risks based on the maximum BAC of the crash participants were nearly as high as those estimated in single-vehicle crashes.