We examined the risk of obstructive respiratory disease associated with tobacco smoke in indoor air, independent of active smoking, ambient air pollution, and some of the other sources of residential indoor air pollution. Data came from a probability sample survey of nine neighborhoods in Philadelphia conducted in 1985-1986, leading to information on approximately 4200 individuals. While for never-smokers the prevalence of obstructive respiratory conditions was proportional to the level of environmental tobacco smoke, this second-hand smoke was not a factor in the frequency of such problems among current smokers. In a series of analyses restricted to never-smokers, each of the 219 index cases of obstructive respiratory disease was matched by age, gender, and neighborhood to three randomly selected controls where matching by neighborhood effectively controlled for ambient air pollution. Both matched and unmatched two-sample analyses showed a statistically significant difference (P = 0.019 and 0.016, respectively) between cases and controls with respect to the level of tobacco smoke in the indoor environment. A conditional logistic regression-matched analysis revealed that heating and cooking as sources of indoor air pollution were not associated with the case/control status. However, the odds ratio for passive smoking at a level of more than one pack per day in the house environment was 1.86 (95% CI, 1.21-2.86). The results show that passive smoking is a significant risk factor for obstructive respiratory disease for never-smokers in an industrialized urban population.