A study of 113 young multiple drug users, interviewed at one year follow-up, was designed to assess the validity of self-reported cannabis use. Subjects provided information about the recency, frequency and typical dose of their use of cannabis in the preceding 30 days. Urine samples were collected and analyzed for delta 9 THC metabolites. Recency of use was the best predictor of urine test result, but a measure of frequency of use significantly increased the proportion of variance accounted for. The results strongly supported the validity of self-reported cannabis use and indicate that valid self-reports of drug consumption can be obtained from multiple drug users in treatment.