Maternal smoking is associated with adverse perinatal outcomes. Because of concerns of underreporting, investigators routinely perform biochemical testing to confirm smoking status, such as serum cotinine, a major metabolite of nicotine, adding an increased cost to examine compliance. The objectives of this study were to determine the sensitivity and specificity of self-reported smoking with serum cotinine as the gold standard and to determine the correlation between self-reported smoking in cigarettes per day and serum cotinine levels. In this cross-sectional study, we surveyed women during the first trimester of pregnancy on their tobacco exposure. A total of 40 women reported that they were smokers, and 40 were nonsmokers, 1 of whom had quit 5 days prior. The mean (+/-S.D.) serum cotinine value among smokers was 155 (+/-122) ng/l, vs. 1 (+/-6) ng/l in nonsmokers, p<0.001. The sensitivity of self-reported smoking status was 97.6%, and the specificity was 100%. Comparing the reported number of cigarettes smoked and the serum cotinine level, the Spearman correlation coefficient was 0.92 (p=0.015) overall and 0.67 (p=0.088) for the subgroup of smokers. This study demonstrates that self-reported smoking exposure during pregnancy is highly accurate. The high correlation coefficient suggests that this is a robust surrogate for cotinine levels.