The study first examines how well statisticians can predict whether and when an individual who has been arrested for a serious violent crime will be arrested again for one of these same crimes. It then considers how well researchers can predict the aggregate number of individuals who, having been arrested for serious violent crimes, will be arrested again for these same kinds of crimes by some time point of interest. The study analyzed all arrests recorded by the Philadelphia police for involvements in violent index crimes by the black and white males in two birth cohorts, one born in 1945 and the other in 1958, who resided in Philadelphia from their 10th through their 18th birthdays and who were arrested at least once for a violent index crime between their 10th and their 27th birthdays. Arrest histories were subdivided into those arrests in the juvenile years and the young adult years. Individual and aggregate sequential-prediction analyses were conducted for each birth cohort, each age interval, and at each successive arrest in the arrest history for violent offenses. The failure time regression models developed show that legally permissible risk variables were not often associated with rearrest risks and timing, that race had a consistent effect except at the first arrest transition, and that few risk variables overall were related to rearrest. Researchers were unable to identify risk variables that were consistently significant across arrest transitions, age groups, and birth cohorts. The study is currently conducting individual- and aggregate-prediction analyses based on these failure time regression findings. 28 tables and 4 figures