BACKGROUND: The trimodal distribution of traumatic death was first described by Trunkey in 1983, which demonstrated that most deaths occur in the first 24 h. We postulate that since 1983, the time-to-death histogram described has shifted. METHODS: A retrospective analysis identifying timing of death was conducted on the National Trauma Data Bank (version 7.2) from 2002 to 2006. Early death was defined as death within 24 h of admission. International Classification of Diseases ninth edition codes with greater than 20% early deaths were called "high-risk codes". Bivariate analyses were conducted to assess the association between demographics, injury factors, and death. Pearson's χ(2) test was used to compare timing of death by region of injury. Multivariate logistic regression was conducted to show the effect of region of injury on death while controlling for demographic factors and injury type. RESULTS: The cohort includes 898,982 patients. The study population was predominantly male (66%) and Caucasian (62%). Mean age and injury severity score were 45 ± 20.3 and 11 ± 10, respectively. Overall mortality rate was 5% with 56% dying early. Head/neck, thorax, and abdomen/pelvis injuries were more prevalent in overall deaths (35%, 22%, and 11%, respectively). Thorax and abdomen/pelvis injuries predicted early death (odds ratio 2.03 and 1.39, respectively). CONCLUSIONS: The prevalence of early death has decreased since 1983, but the majority of deaths still occur within 24 h of injury. Ample opportunity remains to impact mortality in the first 24 h after injury.