Cotinine, a metabolite of nicotine, is widely used to distinguish smokers from nonsmokers in epidemiologic studies and smoking-cessation clinical trials. As the magnitude of secondhand smoke exposure declines because of proportionally fewer smokers and more clean-indoor-air regulations, the optimal cotinine cutpoint with which to distinguish smokers from nonsmokers is expected to change. The authors analyzed data on 3,078 smokers and 13,078 nonsmokers from the National Health and Nutrition Examination Survey for 1999-2004. Optimal serum cotinine concentrations for discriminating smokers from nonsmokers were determined using receiver operator characteristic curve analysis. Optimal cotinine cutpoints were 3.08 ng/mL (sensitivity = 96.3%, specificity = 97.4%) and 2.99 ng/mL (sensitivity = 86.5%, specificity = 93.1%) for adults and adolescents, respectively. Among adults, optimal cutpoints differed by race/ethnicity: They were 5.92 ng/mL, 4.85 ng/mL, and 0.84 ng/mL for non-Hispanic blacks, non-Hispanic whites, and Mexican Americans, respectively. Among adolescents, cutpoints were 2.77 ng/mL, 2.95 ng/mL, and 1.18 ng/mL for non-Hispanic blacks, non-Hispanic whites, and Mexican Americans, respectively. Use of the currently accepted cutpoint of 14 ng/mL overestimates the number of nonsmokers in comparison with the proposed new overall cutpoint of 3 ng/mL or the race/ethnicity-specific cutpoints of 1-6 ng/mL.