A snowball sampling procedure was used to produce the sample, which consisted of persons arrested for drug law offenses, neighborhood residents not currently involved in the criminal justice or social control agency, and participants in a residential drug treatment program. Within each group, participants included persons whose main drug of abuse was crack, cocaine, heroin, and polydrug (primarily marijuana). The confidential and anonymous interviews took place between June 1988 and May 1989. Although differences in initiation into crack and other drugs were expected, results revealed that the social processes of initiation did not differ. However, crack initiates differed from other initiates in their prior involvement with drug abuse and drug selling. Findings indicated that although informal controls continue to influence processes of initiation, the changing social structure of urban areas suggests that these controls will be less influential in maintaining conventions and norms to regulate the use of crack. Thus, our understanding of the processes of initiation into drug use or selling should include new dimensions of what Dembo and colleagues have called "environmental risk." Tables, notes, appended tables, and 101 references (Author abstract modified)