The age distribution of homicide offending in America remained stable until the period 1985 to 1990, when the rates of homicide offending doubled for 15-to-19 year-olds, increased nearly 40 percent for 20-to-24 year-olds, but decreased for those over 30. There have been systematic changes in the age distribution of homicide in the United States associated with “cohort replacement” over the past 40 years. “Cohort replacement” as an explanation for changes in the age distribution of homicide offending suggests that different birth cohorts have characteristics that predispose them to higher or lower propensities for homicide offending. An alternative, popular explanation for the epidemic of youth homicide is the development of crack cocaine drug markets during the last half of the 1980s through the mid-1990s. In testing the validity of these two explanations, this analysis introduces an estimable function approach for estimating the effects of age, period, and birth cohort on homicide offending. This method enables a simultaneous assessment of the impacts of period and cohorts on the age distribution of homicide offending. The analysis found that although the age curve for homicide offending remained relatively stable, there were shifts in the curve systematically associated with cohort replacement, which accounted for nearly half of the upturn in youth homicides during the epidemic period. Two cohort characteristics were linked to much of the variability in homicide offending: relative birth-cohort size and the percentage of nonmarital births; however, a significant fraction of that upturn was not linked to cohort replacement, suggesting the partial effect of the crack cocaine market. 2 tables, 4 figures, and 49 references