No national data have been used to compare self-reported and biochemically assessed cigarette smoking status among adolescents. We investigated discrepancies between self-reported smoking and measurement of serum cotinine concentration among adolescents aged 12-17 years in a representative sample (n=2,107) of the U.S. population. Smoking prevalence was 12.9% among teens who reported in a private interview whether they smoked during the previous 5 days (95% CI=10.9%-14.9%) and 12.5% (95% CI=10.3%-14.7%) according to serum cotinine concentration greater than 11.40 ng/ml (the cutpoint). Among teens who reported being a nonsmoker (i.e., that they did not smoke during the previous 5 days), 2.7% (95% CI=1.6%-3.8%) had a serum cotinine concentration of greater than 11.40 ng/ml. Discrepancies among self-reported nonsmokers were less likely among Mexican Americans than among Whites. Among self-reported smokers, 21.1% (95% CI=13.7%-28.5%) had a serum cotinine concentration of 11.40 ng/ml or less. This discrepancy is explained primarily by the high proportion (37.0%) of teen smokers who reported smoking, on average, less than 1 cigarette per day. We believe that social stigma or fear that their parents would find out about their survey responses may be the main explanation for the 2.7% discrepancy among self-reported nonsmokers, and that smoking patterns (including the extent of nicotine dosing) and a lack of measurement of recency of cigarette smoking are the main explanations for the 21.1% discrepancy among self-reported smokers. Efforts to improve the validity of self-reported cigarette smoking will benefit tobacco-related surveillance, evaluation, and research activities for adolescents.