The author examined the effect of exposure to environmental tobacco smoke (ETS) on hospitalization for respiratory illness in low-birth-weight (LBW, less than 2,500 g) infants. This analysis was performed on the combined data of 3,285 infants from the Jing-An and Chang-Ning epidemiologic studies of children's health in Shanghai. Infants were classified into three ETS groups according to the total number of cigarettes smoked daily by household members: none, light (one to 19 cigarettes/d) and heavy (20+ cigarettes/d). There were no mothers who were smokers. Infants might have been hospitalized more than once during their first 18 mo of life. Risk and incidence density of hospitalization for respiratory disease increased with increasing smoking by household members among the LBW infants more rapidly than among the normal-birth-weight (NBW) infants. Compared with the NBW infants who were living in nonsmoking households, the odds ratios for the first episode of hospitalization for respiratory disease were 1.40 in the NBW infants who were living in light smoking households and 1.61 in those who were living in heavy smoking households. In the LBW infants, the odds ratios were 2.91 and 4.48 in light and heavy smoking households respectively, after adjustment for study area, sex, and feeding by simple logistic regression analysis. Ordinal logistic regression analysis for all episodes of respiratory hospitalization showed similar results.