This study compared the concordance of self-report for recent marijuana use with results obtained from urine drug screen. The sample consisted of adolescent sons of fathers with DSM-III-R lifetime substance use disorder (SUD) [high average risk (HAR); N= 75] and sons of fathers with no Axis 1 psychiatric or SUD [low average risk (LAR); N= 125]. To avoid recall bias, and to ensure that the timeframe for accurate detection was restricted to promote accuracy, urine drug screen results were compared to self-reported marijuana use during the prior 48 h using an interview format. The HAR group of adolescents reported a higher rate of recent cannabis use and also had a higher rate of cannabis detection as measured in urine. The two groups did not differ with respect to either over or under-reporting marijuana use. Overall, of 19 participants who obtained positive urine cannabis results, six (31.5%) verbally denied that they used marijuana within the previous 2 days. Among the 181 participants who obtained a negative urine drug screen, 20 subjects (11.5%) inaccurately asserted that they had used marijuana within the prior 2 days. In total, 13% of the participants (26/200) inaccurately reported recent cannabis use. These results underscore the need for caution in interpreting the results of self-report methods documenting the prevalence of drug use among youth.