Study participants were 178 male, drug-using inmates of two Dutch penitentiaries. In a prospective study, discriminant analysis was used to explore which variables distinguished between participants with and without violent crime during a 2-year follow-up period after detention. The study assessed predictive validity of demographic, criminal history and ASI variables. Overall correct classification of violent crime varied between 82 percent and 93 percent. Apart from the ASI severity rating for criminal past, the number of years of regular cocaine use and the age at which cocaine use started were major contributors of the ASI’s predictive power. The ASI-based assessment of drug use can indicate a risk of violent crime. Further research is needed into the processes involved in this predictive power, as well as into the circumstances in which drug use induces violent crime. Tables, references