This study examined the offending of brothers and sisters of the males in the Cambridge Study in Delinquent Development, which involves a longitudinal survey of the development of offending and antisocial behavior of 411 males who were first contacted in 1961-62. The research examined how effective risk factors were in predicting the offending of these brothers and sisters. It also examined gender differences in risk factors and their implications for prevention strategies. The study found that the number of convictions for criminal offenses was much higher for brothers (44 percent) than for sisters (12 percent). Offense types varied between brothers and sisters. Burglary and theft of vehicles were more prevalent for brothers; whereas, shoplifting and deception offenses predominated for sisters. Offending tended to be concentrated in particular families, and the probability of a child being convicted for an offense increased with the number of other children in the family who had been convicted. The important risk factors for offending were similar for brothers and sisters; however, there were some gender differences. Socioeconomic and child-rearing factors ( e.g., low family income and poor parental supervision) were more important for sisters; and parental characteristics (e.g., nervous fathers and mothers) were more predictive of offending for brothers. Risk scores predicted offending more accurately for sisters than for brothers. Existing theories about gender differences in offending are not confirmed in this research, so new theories are required to help explain these study findings. The findings suggest that parent training and parent education techniques that target parental discipline, supervision, reinforcement of positive child behavior, and increasing parental interest in children are more likely to reduce female offending than male offending. Still, the absolute number of offenses reduced by these measures is likely to be greater for males than for females due to the higher prevalence of male offending. 3 tables and 3 references