To the Editor:Hajek and Snuggs said that our evidence of the validity of smoking self-reports was uninformative because smokers probably opted out of providing blood samples to avoid being forced to admit that they smoked.But in fact, in the NHANES surveys we analyzed, respondents were given a list of the tests that would be performed with their blood samples, and detecting cigarette consumption was not listed (http://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/examinationconsent_0304.pdf).Furthermore, according to an official from the Centers for Disease Control and Prevention, who runs the surveys, “participants do not know that we will be testing their blood for cotinine levels or for exposure to environmental tobacco smoke” (D.J. Brody, personal communication, January 9, 2011). Therefore, it is unlikely that some smokers declined to provide a blood test because they did not want their smoking to be detected.Hajek and Snuggs cited 2 studies1,2 purportedly showing high rates of under-reporting of smoking among pregnant women who were not aware that their blood samples would be compared with their self-reports. But these studies are atypical as many other studies of pregnant women documented low levels of misreporting. For example, Derauf et al3 found that only about 3% of pregnant women manifested elevated levels of cotinine in their babies’ meconium yet claimed not to smoke cigarettes during the routine patient intake questionnaire administered immediately before delivering the baby. These women were not told that their meconium would be compared with their self-reports.Similarly, Klebanoff et al4,5 compared cotinine levels in blood samples taken during a routine OB examination in the first trimester (not described as part of a study of smoking behavior) with self-reports provided at the same time and found that only about 5% of the women said they were not smokers yet manifested elevated cotinine levels. Similar findings have been reported by numerous other investigators.6–8 Based on such evidence, Klebanoff et al5 concluded that in many circumstances measurements of smoking “are not likely to differ substantially whether based on self-report or cotinine.”The prevalence of underreporting in these studies of pregnant women is very similar to what we found when we compared self-reports of cigarette smoking from American general public samples with the results of contemporaneous blood tests: 5.2%. Furthermore, the NHANES data documented that this figure is inflated due to the use of other products containing nicotine. Taking into account passive smoking as well, the rate of intentional misreporting appeared to be near zero.A likely reason why Ford et al1 and Pärna et al2 found higher rates of mismatching is that they compared the results of blood tests with women's reports made months later of whether they had been smoking at the time of the blood tests. In other words, these studies assessed the accuracy of long-term recollections of smoking status, not the accuracy of reports of smoking status at the time the reports were provided. Thus, the apparent inaccuracy of the self-reports could be the result of intentional misrepresentation, or it could be the result of unintentional misremembering instead.Finally, consider experiments in which some adults (selected randomly) provided self-reports of their smoking behavior after providing physiological samples that they were told will reveal their cigarette consumption, and other adults provided self-reports and were not asked to give physiological samples. Rates of self-reported smoking were the same in both groups in such experiments,9 suggesting that adult smokers do not lie when providing self-reports.