After relying on a review of historical research, a longitudinal dataset, and multivariate analyses in an analysis of drug-violence relationship, a proposed integrated model approach of chronic violent offending and the effect of drug use were studied. The model was intended to predict frequency of serious violent offending, while controlling for simultaneous drug use and other known theoretical constructs that are consistently related to violence. The data used were primarily from the first two waves of the National Youth Survey (NYS) collected in 1976 and 1977. The results suggest that the predictors of violence are dynamic and complex. Drug use was an important predictor of violence, but the nature and direction of the relationship changed. The study supports some methods for facilitating desistance from violent offending, particularly intervention/prevention efforts that attempt to revitalize dilapidated neighborhoods and enforcement efforts that disrupt youth involvement in drug markets. Tables, references