The “victim labeling” hypothesis suggests that the experience of prior victimization increases victimization in the future. The “victim rationality” hypothesis suggests that past victimization decreases subsequent victimization due to lessons learned from past victimization. Results from the current analyses provide support for the “victim labeling” hypothesis, i.e., past victimization increased subsequent victimization; however, the magnitude of this link was more modest than has been found from analytical models used in prior longitudinal victimization research. Study data came from the Rural Substance Abuse and Violence Project, a prospective longitudinal study of substance use, criminal victimization, and criminal offending among a panel of students in Kentucky. Data were initially collected in the spring of 2001 when the panel was in the seventh grade; followup surveys were conducted during the spring of the three subsequent years (2002, 2003, and 2004). The analyses are based upon a subset of 2,706 individuals from the larger sample. This sample constitutes the maximum number of individuals with complete data for three adjacent time points. The dependent variable was a self-report of the number of times that the respondent was physically attacked on school grounds or during school-related activities during the current school year. Independent variables were prior-year victimization; personality characteristics; prosocial ties and activities; and risky places, peers, and behaviors. Recommendations are offered for the design of future research in this domain. 3 tables, 3 appendices, and 74 references