The prevalence of cannabis use is rising among adolescents, many of whom perceive little risk from cannabis. However, clinicians who treat adolescent substance users hear frequent reports of serious cannabis-use disorders and problems. This study asked whether cannabis produced dependence and withdrawal among such patients, and whether patients' reports supported previous laboratory findings of reinforcing effects from cannabis. This was a screening and diagnostic study of serial treatment admissions. The diagnostic standard was the DSM-III-R dependence criteria, and the setting was a university-based adolescent substance treatment program with male residential and female outpatient services. The patients were 165 males and 64 females from consecutive samples of 255 male and 85 female 13-19-year-olds referred for substance and conduct problems (usually from social service or criminal justice agencies). Eighty-seven patients were not evaluated, usually due to early elopement. Twenty-four others did not meet study admission criteria: > or = one dependence diagnoses and > or = three lifetime conduct-disorder symptoms. The main measures were items from diagnostic interview instruments for substance dependence, psychiatric disorders, and patterns of substance use. Diagnoses were substance dependence, 100%; current conduct disorder, 82.1%; major depression, 17.5%; attention-deficit/hyperactivity disorder, 14.8%. The results show that most patients claimed serious problems from cannabis, and 78.6% met standard adult criteria for cannabis dependence. Two-thirds of cannabis-dependent patients reported withdrawal. Progression from first to regular cannabis use was as rapid as tobacco progression, and more rapid than that of alcohol, suggesting that cannabis is a reinforcer. The data indicate that for adolescents with conduct problems cannabis use is not benign, and that the drug potently reinforces cannabis-taking, producing both dependence and withdrawal. However, findings from this severely affected clinical population should not be generalized broadly to all other adolescents.