Demonstration of a dose-response relationship for environmental tobacco smoke (ETS) is an important indication of causality. Central to the analysis and interpretation of dose-response relations as described in epidemiological studies is the relationship between dose and exposure. It must be recognized that in studies of ETS we have only surrogate measures of dose, and these surrogate measures (based on exposure) are imperfect. The question-based measures of ETS exposure generally have not been standardized, may have limited validity and reliability, and cannot comprehensively describe total ETS exposure, exposure to individual ETS components, nor doses of biologically relevant agents at target sites. Nevertheless, useful data have been yielded in epidemiologic studies linking ETS exposure to increased respiratory infection and symptoms, reduced lung growth in children, and increased lung cancer in nonsmoking adults. The more consistent exposure-response data for studies on acute health in children may reflect the greater difficulty in measuring exposure in studies of chronic health in adults.