This research examined the validity of self-reports of adolescent smoking and explored factors that may influence agreement between self-reported smoking and biological indicators. Data were obtained from 1,881 adolescents as part of a household probability study in the southeastern United States. Adolescents aged 12-14 years reported their tobacco use and provided breath and saliva samples. The sensitivity and specificity of self-reports of smoking were calculated with measures of carbon monoxide, salivary cotinine, and salivary thiocyanate as standards. When the self-report items were compatible with the half-life of the biological indicator, and values were adjusted for smokeless tobacco use, sensitivity varied widely (carbon monoxide, 38%; cotinine, 81%-87%; thiocyanate, 80%). The range for specificity was more constrained (carbon monoxide, 98%; cotinine, 98%; thiocyanate, 80%). Sensitivity of self-report was influenced by the biological indicator examined and the timeframe of questions on smoking. Specificity of self-report was influenced by the biological indicator examined and use of other tobacco products. Overall, sensitivity and specificity were highest when recent smoking was assessed and when cotinine was the standard and smokeless tobacco use was considered. Examination of inhalation patterns suggested that detection of smoking may be influenced by recency of smoking and whether the adolescent smoker inhales. Assessments of individual adolescents' smoking were most accurate when smoking was recent and when cotinine was used and smokeless tobacco could be identified. Prevalence estimates of smoking were similar with self-reports and cotinine.