Differences in baseline characteristics between smoking categories were compared with Student t tests or Mann-Whitney test for continuous variables and χ2 tests for categorical variables. The associations among period of maternal smoking during pregnancy (no, first trimester only, continued), paternal smoking (no, yes), and the reported number of cigarettes smoked with wheezing in children at ages 1, 2, 3, and 4 years were analyzed using multiple logistic regression models. The associations of paternal smoking (no, yes) with wheezing in children at age 1, 2, 3, and 4 years were performed among mothers who did not smoke during pregnancy (n = 3,468). Using similar models, we assessed the associations with the number of wheezing episodes. All models were adjusted for potential confounders, including parental age, education, ancestry, maternal parity, children’s sex, gestational age, birth weight, familial history of asthma and allergy, breast-feeding, day-care attendance, and pet keeping. Confounders were included in the models based on their associations with wheezing or a change in effect estimate of > 10%. Therefore, parity and paternal age were not included in the models. Finally, to assess whether the associations of maternal smoking during pregnancy with wheezing were modified by postnatal smoke exposure and gestational age-adjusted birth weight, we repeated these analyses with wheezing as the outcome in strata of postnatal smoke exposure and small size for gestational age, defined as the lowest 10% of gestational age-adjusted birth weight in the cohort. Tests for trends were performed by treating each categorized variable as a continuous term and entering the variable into the fully adjusted regression model. Missing data in the covariates were imputed using the multiple imputation procedure to select the most likely value for a missing response. Five imputed data sets were created and analyzed together. No major differences in the magnitude or direction of the effect estimates were observed between analyses with imputed missing data and complete cases only. The maximum change in ORs for ever wheezing was 6%. We only present the results based on imputed data sets. All measures of association are presented with their 95% CIs. Statistical analyses were performed using SPSS version 17.0 for Windows software (SPSS Inc).