Summary Objectives To compare smoking prevalence estimates from two nationally representative surveys in the USA. Study design: Smoking prevalence estimates derived from the National Health Interview Survey (NHIS) and the National Survey on Drug Use and Health (NSDUH) for the period 1998–2005. Methods Comparisons according to age (18–34 or 35+ years) and according to smoking frequency (every day or some days). Results In 1998, the prevalence of smoking in both surveys was nearly identical at 24%. From 1999 to 2005, a divergence occurred in smoking prevalence found by the NSDUH and the NHIS. By 2005, NHIS prevalence had declined to 20.9% [95% confidence interval (CI) 20.3–21.5], but the NSDUH estimate was 25.4% (95%CI 24.6–26.2), amounting to 9.1 million more smokers. In 1999, prevalence among 18–34 year olds in the NSDUH was only 18% (95%CI 13–22) higher than that in the NHIS, but that difference had doubled by 2005, when smoking prevalence among 18–34 year olds was 36% (95%CI 30–41) higher in the NSDUH than in the NHIS. NSDUH and NHIS prevalence among 35+ year olds were similar in 1999 and 2001, but the difference was 13% (95%CI 9–18) in 2005. Higher smoking prevalence estimates in the NSDUH were largely due to much higher estimates for some-day smoking in that survey, although every-day smoking prevalence among 18–34 year-olds was also higher in the NSDUH than in the NHIS. Conclusions These results raise doubt about the recent decline in smoking prevalence described by the NHIS. Further investigation of the NSDUH/NHIS discrepancy may lead to better surveys and to a clearer picture of smoking trends in the USA.