The study involved a nonprobability sample of youth who exhibited mental health and substance-abuse problems and had contact with the juvenile justice systems in 1 of 11 counties in New York State. Data on each youth included information routinely collected by probation officers during intake interviews; accompanying medical, psychiatric, and school records; and historical juvenile records. All youth in the sample had been referred to the diversion program since 1997. Violent offending was measured as a dichotomous variable and included assault (simple and aggravated), rape and other sexual assaults, and robbery. Potentially violent offenses, such as aggravated harassment and menacing behavior, were also included. The findings provide mixed support for the use of mental-health and substance-abuse history as predictors of violent offending for juveniles. Mental-health history was a significant predictor of later violent offending in the main effects model, indicating the importance of properly treating those youth who are referred for mental health treatment in childhood and early adolescence. Also, juveniles who commit violent offenses or manifest problem behavior during this developmental stage should be evaluated for possible mental health disorders. Substance-abuse history actually reduced the odds of being charged with a violent offense. This may be due to the fact that most juveniles in the sample who used drugs were using marijuana, which has not yet been conclusively linked to violent offending. The interaction between prior mental health evaluation and substance-abuse history, however, was a significant predictor of violent offending. The authors advise that an understanding of the complex relationships among a number of different risk factors is essential in developing and implementing strategies that effectively prevent violent behavior during adolescence and into adulthood. 5 tables and 30 references