To determine differences between racial/ethnic groups in overall rates of death by homicide, proportion of firearm homicides, and the use of cocaine prior to death. Descriptive epidemiologic survey of a complete 2-year sample of homicides. New York City, NY (population 7,322,564). All residents (N = 4298) of New York City who were victims of homicide during 1990 and 1991. Using medical examiner data, age- and gender-specific rates of homicide were calculated for African Americans, Latinos, and whites. Separate logistic regression analyses were conducted to examine the association between demographic variables and both recent cocaine use and firearm-related homicides. Young African-American and Latino men were more likely to be victims of homicide than all other demographic groups. Approximately three fourths of all homicides involved firearms. In the subset of homicide victims dying within 48 hours (n = 3890), 31.0% were positive for cocaine metabolites. African Americans (odds ratio [OR], 1.6; 95% confidence interval [CI], 1.2 to 2.1), Latinos (OR, 1.4; 95% CI, 1.1 to 1.9), and victims 25 through 34 years of age (OR, 2.9; 95% CI, 2.5 to 3.5) and 35 through 44 years of age (OR, 2.7; 95% CI, 2.2 to 3.4) were more likely to be positive for cocaine metabolites than other groups. Young males and females were equally likely to have used cocaine before death. Victims 15 through 24 years of age were more likely than other age groups to be killed by a firearm. African Americans (OR, 1.7; 95% CI, 1.3 to 2.3), Latinos (OR, 1.5; 95% CI, 1.2 to 2.0), and Asians (OR, 2.2; 95% CI, 1.4 to 3.6) were more likely than whites to be killed by a firearm. Men (OR, 4.8; 95% CI, 4.0 to 5.9) were more likely than women to be killed by a firearm. There was no association between having used cocaine before death and being killed by a firearm. The high rates of death by homicide among young African Americans and Latinos may be due to the increased involvement with both cocaine use and firearms. New efforts must be made to decrease cocaine use and firearm availability.