The study sample (n=612) was derived from the original subjects (approximately 8,000) of the National Collaborative Perinatal Project in Philadelphia. This project began in the late 1950's. The sample for this study consisted of an equal number of male and female African-Americans with an average age of 26.23. Compared with national norms for Black adults, the study sample had elevated rates of drug use for most of the drug categories. Seventy-three percent of the sample reported lifetime marijuana use, compared with 37 percent for the national norm for Black adults; this pattern was true for both sexes. The instrument used to assign subjects to either delinquent or nondelinquent subgroups was the "Peer Activity and Culture" Section of the Texas FFS (Family, Friends, and Self Scale). The frequency of drug use and the duration of "regular" drug use were measured for "lifetime" up to age 26. "Regular" use was defined as use at least 2 days per week. For use as predictor variables in the analyses for determining the relationship of the degree of use of each type of illicit drug to the degree of violent behavior index, scores were calculated for marijuana, sedatives/tranquilizers, and cocaine/crack. Two control variables were used to correct partially for the limitations of self-report data: The Minnesota Multiphasic Personality Inventory Lie Scale score and the Crowne-Marlowe Social Desirability Scale. The dependent outcome variables were measures of illegal, violent behavior. The PPC Delinquency and Criminal Behavior Inventory, administered in a structured interview format, was used to identify each of 20 types of illegal offenses, as well as the number of convictions for each type of offense. For the conventional, nondelinquent subgroup, a higher degree of significant relationship between degree of marijuana use and degree of violent behavior was found, compared to the degree of this type of relationship for either cocaine/crack use, amphetamine use, or tranquilizer/sedative use. Thus, a special disinhibition effect was found only for marijuana. No evidence is presented on the process by which marijuana use predisposes youth to violent behavior; however, the authors theorize that the marijuana-violence linkage stems from a socioeconomic, environmental, and lifestyle influence, rather than from a biochemical or neuro-physiological effect. 5 tables and 17 references