The joint impact of maternal race/ethnicity and age on the relationship between smoking and hypertensive disorders of pregnancy. The present study found that the association between maternal cigarette use and hypertensive disorders of pregnancy varied by maternal race/ethnicity and age. Specifically, the decreased odds of PIH among women who smoked during pregnancy was only WZ8040 apparent in non-Hispanic white and American Indian women younger than 35 years old who smoked during pregnancy, based on the natality data and among non-Hispanic white only based on the NIS sample. Interestingly, we also observed that maternal cigarette use during pregnancy was associated with increased odds of PIH for non-Hispanic Asians regardless of maternal age, based on the natality data and conferred for the older age group by the NIS data. In general, we observed an association in the same direction with cigarette use during pregnancy and PIH between the two data sets, albeit with differences in the strength of association with a weaker association from the NIS data for women younger than 35 years of age, but a stronger association for mothers with age greater than or equal to 35. This discrepancy in findings between the two data sets may be explained by the different sources of information between the two data sets. In the NIS data, measures of interest were defined based on ICD-9 discharge diagnosis codes, whereas in the natality data smoking is selfreported by the patient and hypertensive disorders were coded by chart extractors. Hence, compared to the natality data, prevalence of maternal smoking during pregnancy in the NIS data was much lower, whereas, the prevalence of the outcome was higher. Geller and colleagues evaluated the accuracy of the ICD-9 revision codes for preeclampsia and eclampsia and observed variation in accuracy of diagnosis with a positive predictive value for severe preeclampsia of 84.8%, 45.3% for mild preeclampsia, and 41.7% for eclampsia. The potential misclassification in NIS data for exposure and the outcome were likely non-differential, however, which would bias the point estimate toward the null value and may explain the weaker strength of association observed in the younger women in the NIS data. The inverse association between maternal smoking and PIH among non-Hispanic white and American Indian women in our study appears to be weaker compared to findings from prior research. The weaker association between maternal smoking and PIH from our study may be attributed to potential misclassification of the exposure, maternal smoking, due to recall bias or underreporting for both natality and the NIS data. Nevertheless, this measurement error of exposure was likely non-differential.