About This Report

Fatal crash data from FARS and nonfatal crash data from GES are presented in this report in five chapters. Chapter 1, “Trends,” presents data from all years of FARS (1975 through 2004) and GES (1988 through 2004). The remaining chapters present data only from 2004. Chapter 2, “Crashes,” describes general characteristics of crashes, such as when and how often they occurred, where they occurred, and what happened during the crash. Chapter 3, “Vehicles,” concentrates on the types of vehicles involved in crashes and the damage to the vehicles. Chapter 4, “People,” is the largest chapter of this report, with statistics about drivers, passengers, pedestrians, and pedalcyclists. The last chapter of the report, “States,” contains information about crashes for each state, the District of Columbia, and Puerto Rico. Terms used throughout the report are defined in the Glossary.

About three-quarters of the tables in this report present data from both FARS and GES. The remaining tables contain FARS data only. Statistics describing fatal crashes or fatalities have been derived from FARS. Statistics describing injury crashes, property-damage-only crashes, or nonfatal injuries have been derived from GES. The reader should be aware that FARS numbers are actual counts of fatalities or fatal crashes, whereas GES numbers are estimates of counts of crashes and injuries and are subject to sampling and nonsampling errors. (See Appendix C for more information on these errors.) To emphasize this difference, FARS numbers are not rounded, while GES estimates have been rounded to the nearest thousand. As a result of the rounding, for some tables, the sum of the row or column entries may not equal the row or column total. In addition, percentages have been calculated prior to rounding.

The reader may also notice that many tables have rows or footnotes for “unknowns” for FARS data, but not for GES data. The reason for this difference is that almost all the GES unknown data have been assigned values through complex statistical procedures. FARS unknown data, on the other hand, are not assigned values, with the exception of blood alcohol concentration (BAC) test results. When the alcohol test results are unknown, BAC values have been assigned to drivers and nonoccupants involved in fatal crashes, using a method of multiple imputation that was revised in 2001. More information on the new multiple imputation method, including detailed tabulations of alcohol involvement in various categories (age, sex, time of day, etc.), is available in NHTSA Technical Report DOT HS 809 403, Transitioning to Multiple Imputation: A New Method to Estimate Missing Blood Alcohol Concentration (BAC) Values in FARS.