Although cigarette yields of tar and nicotine have been declining since the early 1970s, little information is available for the general population on the consequences of their use on exposure to tobacco combustion products. In a population-based sample of 298 smokers, the majority of whom were Hispanic, we examined the relationships between yields of cigarettes currently smoked and levels of salivary cotinine and end-expired carbon monoxide. Spearman correlation coefficients between the current number of cigarettes smoked and cotinine (r = 0.52) or carbon monoxide (r = 0.51) were higher than correlations between the Federal Trade Commission (FTC) nicotine data and these same markers, 0.12 and 0.05, respectively. Correlations between FTC tar and carbon monoxide yields and the biologic markers were similarly weak. In multiple linear regression models, the current number of cigarettes smoked was the most important predictor of cotinine and carbon monoxide levels (p < 0.0001). The addition of FTC tar, nicotine, or carbon monoxide to the models explained little of the variability in cotinine or carbon monoxide levels. Because FTC yields of tar and nicotine are poor predictors of exposure to tobacco combustion products, subjects' reports of cigarette brand should not be used as a primary marker of exposure in epidemiologic investigations. Furthermore, smokers need to be advised about the limitations of cigarette yield information for predicting the potential for adverse health effects of smoking.