The effects of household exposure to cigarette smoke on hospitalization and incidence of respiratory illness were examined among 2227 children at Chang-Ning District, Shanghai Municipality, People's Republic of China. The passive smoking quantity was estimated by total daily cigarette consumption of family members and number of cigarettes smoked in the home. No mothers who smoked were found. A significant dose-response relationship of passive smoking to hospitalization for respiratory illness during the children's first 18 months of life was found, for which no confounding factors were discovered. The incidence density ratio of hospitalization for respiratory illness was 2.1 for children living in families including people who smoked 20 or more cigarettes a day compared with those living in non-smoking families. The children appeared to be more vulnerable in the first six months of life than in the 7-18 month period, and those with lower birth weight and the artificially fed were more susceptible. The cumulative incidence of bronchitis or pneumonia increased significantly with increasing cigarette smoking of family members, which persisted when sex, birthweight, nursery care, father's education, coal for cooking, and adult cases with chronic respiratory disease were taken into account. Family smoking status was not found to be significantly associated with the incidence of asthma, whooping cough, sinusitis and measles.