A large study was conducted to assess exposure to environmental tobacco smoke (ETS) in a geographically dispersed study population using personal breathing zone air sampling and salivary cotinine levels. Approximately 100 self-reported nonsmoking subjects in each of 16 metropolitan areas were recruited for this investigation. Cumulative distributions of salivary cotinine levels for subjects in smoking and nonsmoking homes and workplaces exhibited a general trend of decreasing salivary cotinine levels with decreasing time spent in smoking environments. Median salivary cotinine levels for the four experimental cells in the study (product of smoking and nonsmoking home and workplaces) were comparable to those reported for a large national study of serum levels of cotinine (Third National Health and Nutrition Examination Survey, NHANES III), when the latter was corrected for expected differences between serum and saliva concentrations. However, the most highly exposed group in this study had a median salivary cotinine concentration approximately a factor of 2 greater than that of the comparable group in the NHANES III study. Misclassification rates, both simple (for self-reported nonsmokers) and complex (self-reported lifetime never smokers), were near the median of those reported for other studies. Estimated misclassification rates for self-reported lifetime never-smoking females are sufficiently high (2.95% using a discrimination level of 106 ng/ml) that, if used in the Environmental Protection Agency (EPA) risk assessment related to ETS and lung cancer, would place the lower 90% confidence interval (CI) for relative risk at nearly 1.00, i.e., no statistically significant increased risk. For the 263 most highly exposed subjects in the study whose self-reported nonsmoking status was accurate, the correlation between airborne exposure to nicotine and average salivary cotinine is so small, on an individual basis, that it makes the relationship useless for estimating exposure on a quantitative basis. When subjects are grouped according to likely categories of nicotine exposure, correlation between group median airborne nicotine exposure and salivary cotinine level increases dramatically. The comparison improves for the most highly exposed subjects, suggesting that such quantitative comparisons are useful for only those subjects who are exposed to the higher levels of ETS. However, airborne nicotine exposure for most of the subjects does not account for estimated systemic levels of nicotine, based on salivary cotinine levels.