Results indicate that individuals from the three samples experienced different risks for lethal victimization due to their different demographic and compositional characteristics. In particular, Black individuals and those released into Los Angeles, CA had significantly increased risks of homicide victimization. One cohort effect was found in the data for the Hispanic members of the 1991-1992 sample. In the two earlier samples, the homicide mortality differences between Whites and Hispanics were not significantly different, but for the 1991-1992 sample, Hispanics had significantly elevated hazard rates of homicide compared to Whites; in the 1991-1992 sample, Hispanics had hazard rates that were statistically consistent with Black members of the sample. Family criminality was the only family lifestyle variable that significantly predicted homicide victimization risk. Gang activity had a demonstrably large effect on the risk of homicide victimization, and this gang effect appeared largely responsible for the increased homicide rate in the1991-1992 sample. Also found was that the release from a period of incarceration and violent offending significantly increased the risk of homicide victimization, but the level of overall offending did not influence the homicide risk. Other behavioral problem measures, such as school dropout status and evidence of prior drug and alcohol abuse, failed to predict the risk of homicide victimization. Data were collected from a pooled analysis on three random samples of male California Youth Authority (CYA) wards. Tables, figures, appendix, notes, and references