Data for this study were drawn from arrestees in 23 urban areas from the NIJ Arrestee Drug Abuse Monitoring (ADAM) program. The study included interviews using a self-report questionnaire and a urine sample for drug testing. Testing positive for cocaine use upon arrest decreases the probability that the current arrest is for a violent crime, across categories of gender and race, with the relationship particularly strong for females. These results counter psychopharmacological explanations of the cocaine-violence relationship. Factors related to both cocaine use and violent offending may combine to appear as a spurious relationship between them. In addition, the relationship between cocaine and violent offending may vary across groups: age, education and employment status, gender, race, marital status, and generalized deviance and substance abuse. The article observes that cocaine use appears to be conducive to property offending, and should be targeted for its role in spurring or facilitating those kinds of offenses. In addition, cocaine use may lead to violence in connection with the need for money to sustain the habit. Notes, tables, references