The reported association between passive smoking and respiratory illness in children has been based on the parents' assessment of their own level of smoking. To more critically evaluate a causal relationship between passive smoking and childhood ill health, we used urinary cotinine, which is the major metabolite of nicotine and has a long half-life, to objectively quantitate the level of passive smoking in children. Urine was collected from 609 children (median age 3.8 yr, range 1 month to 17 yr) on admission to hospital; cotinine levels were obtained in 491 of these samples, and a comprehensive respiratory questionnaire was completed for 468 children. Statistical analysis was carried out on transformed data using both parametric and nonparametric statistics. Cotinine levels in the children correlated with the parents' current smoking (p less than 0.001). Elevated levels were found in the 41 children admitted with bronchiolitis compared with a group of a similarly aged children with nonrespiratory illnesses (p less than 0.02). Elevated levels were not found for any other diagnosis. We conclude that the urinary cotinine approach has provided objective evidence linking passive smoking to hospital admission for bronchiolitis in infants.