In this study, I examine the extent to which the presence of delinquent peers influences students’ own delinquency risk by using large administrative data from North Carolina public high schools and state government records. Fixed effect regression estimates, which exploit year-to-year, within-school variation in the share of former delinquents in the ninth- grade cohort, show that having more delinquent high school peers, especially those with similar demographic characteristics, tends to increase a ninth grader’s own delinquency risk. A separate analysis of co-offenses committed with formerly delinquent ninth-grade peers shows that co-offending is a potentially important mechanism of delinquency spillovers. I conduct a series of robustness checks and find that the main findings remain robust across alternative specifications.