The United States Department of Defense medical planners need survival-time estimates for anticipated patient streams associated with projected combat scenarios. Survival-time estimates should be grounded in empirical observations. Unfortunately, research in this domain has been limited to a single paper describing the development of died-of-wounds curves for combat casualties with life-threatening injuries. The curves developed from that research were based on a small dataset (n = 160, with 26 deaths and 134 survivors) of forward surgical (Role II) casualties and subject matter experts' judgments. This paper reports the first empirically based time-to-death curves for combat casualties based on a large sample. The results indicate that survival time varied across roles of care at which casualties died but was at most weakly associated with injury severity. Time-to-death curves were, therefore, developed for the overall study population of valid times to death and for Role I, Role II and Role III care. The log-logistic probability distribution provided the best representation of the survival times for the overall study population, while the log-normal distribution was the best choice for Role I, Role II and Role III care. The proposed time-to-death curves should refine the survival-time estimates used in combat medical logistics planning.