Although clinic-based studies have used biochemical validation to estimate the percentage of pregnant women who deny smoking but are actually smokers, a population-based estimate of nondisclosure of smoking status in US pregnant women has not been calculated. The authors analyzed data from the 1999-2006 National Health and Nutrition Examination Survey and estimated the percentage of 994 pregnant and 3,203 nonpregnant women 20-44 years of age who did not report smoking but had serum cotinine levels that exceeded the defined cut point for active smoking (nondisclosure). Active smoking was defined as self-reporting smoking or having a serum cotinine concentration that exceeded the cut point for active smoking. Overall, 13.0% (95% confidence interval (CI): 8.8, 17.1) of pregnant women and 29.7% (95% CI: 27.3, 32.1) of nonpregnant women were active smokers. Nondisclosure was higher among pregnant active smokers (22.9%, 95% CI: 11.8, 34.6) than among nonpregnant smokers (9.2%, 95% CI: 7.1, 11.2). Among pregnant active smokers, nondisclosure was associated with younger age (20-24 years). Among nonpregnant active smokers, nondisclosure was associated with Mexican-American and non-Hispanic black race/ethnicity. Studies and surveillance systems that rely on self-reported smoking status are subject to underestimation of smoking prevalence, especially among pregnant women, and underreporting may vary by demographic characteristics.