This study examines the question of whether alcohol or drug use causes teenagers to engage in violent behaviors as measured by physical fighting, carrying a gun or carrying other types of weapons. The study used two-stage least squares estimates to establish causality. The method first predicted consumption using the prices of beer, marijuana, and cocaine--and then entered predicted consumption in the violence equation. This technique allowed the consumption measures to be purged of their correlation with unobserved characteristics. Data came from the National School-Based Youth Risk Behavior Study, which contains nationally representative samples of high school students. Beer and marijuana consumption did cause teens to engage in more physical fights, while cocaine use appeared to have had no relationship. None of the substances led to increased probabilities of carrying a gun or other weapon. References, notes, tables, appendix