The effects of passive tobacco smoke exposure upon respiratory symptoms and lung function were assessed in a cross-sectional survey of 770 children 7 yr of age, using cotinine as a quantitative biochemical marker of exposure. Salivary cotinine levels were strongly related to the number of smokers in the home, but three-quarters of children from nonsmoking households had detectable salivary cotinine, and 10% of this group were in the upper two-fifths of the distribution of measured tobacco smoke exposure. Smoking by persons other than members of the household may deserve greater attention in future studies of young children. After adjustment for housing tenure, most respiratory symptoms were unrelated to salivary cotinine, but a "tendency for colds to go to the chest" was twice as prevalent in the upper two-fifths as in the lower two-fifths of the cotinine distribution. No association was found between salivary cotinine and reports of wheeze or measured reduction in FEV1 after 6 min of free running. After adjustment for sex, height, test conditions, and housing tenure, all baseline spirometric indices except FVC were inversely associated with salivary cotinine. Only FEF75-85 and FEF75 were significantly reduced, the difference for each index between the top and bottom quintiles of the cotinine distribution being about 7%, equivalent to a reduction of 1.1% (95% CL, 0.1 to 2.1%) per doubling of cotinine concentration. These changes may be evidence of small airways damage, which could later progress to more severe respiratory impairment.