Assessment of self-reported smoking behavior in cardiovascular studies may lead to inaccurate measures of nicotine exposure. A more objective measurement of nicotine exposure can be done by measurement of plasma cotinine levels. The aim of the present study was to define the rate of discordance between the self-reported smoking behavior and biochemically defined smoking status. Data from 3,316 patients hospitalized for coronary angiography, who completed a questionnaire on smoking behavior, were analyzed. As a biochemical assessment of smoking status we used a cut-off serum cotinine level of 15 μg/l. Smoking denial, defined as a discrepancy between high cotinine levels and self-reported never- or ex-smoking status, was observed in 3.7 % of the study participants. In a logistic regression analysis with a step-wise inclusion of sex, age, CAD, previous MI, and educational level, only male sex (odds ratio male/female: 2.00, 95 % CI 1.22-3.33; p = 0.007) and age (odds ratio per year: 0.79, 95 % confidence interval 0.66-0.94, p = 0.008) were associated with smoking denial. In conclusion, a misclassification rate of 3.7 % in the evaluation of such an important risk factor may lead to blurred effects and favor false negative results. The results of the present study substantiate the importance of biochemical markers for smoking assessment in cardiovascular studies.