The distribution of time from acute traumatic injury to death has three peaks: immediate (less than or equal to one hour), early (6 to 24 hours), and late (days to weeks). It has been suggested that coordinated trauma care dampens the late peak; however, this research may be more reflective of unintentional than intentional deaths. This study examines whether a coordinated trauma system (TS) alters the temporal distribution for assault-related deaths. Data were obtained from homicides examined by the Jefferson County Coroner's/Medical Examiner's Office from 1987 to 2008. Homicides were categorized-based on year of death-as occurring in the presence of no TS, during TS implementation, in the early years of the TS, or in a mature TS. The temporal distribution of homicide mortality was compared among TS categories using a χ(2) test. A Cox Markov multistate model was used to estimate proportional changes in the temporal distribution of death adjusted for assault mechanism. With a TS, after adjusting for assault mechanism, a lower proportion of homicide victims survived through the first hour (hazard ratio [HR], 0.75; 95% confidence interval [CI], 0.54 to 1.03) and from one to six hours (HR, 0.68; 95% CI, 0.49 to 0.96). Additionally, the presence of a TS was associated with a proportional decrease in deaths after 24 hours (P = 0.0005). These results suggest that a trauma system is effective in preventing late homicide deaths; however, other means of preventing death (such as violence prevention programs) are needed to decrease the burden of immediate homicide-related deaths.