The measurement of bronchial hyperreactivity (BHR) may give additional information to questionnaire-based studies of asthma prevalence. It is desirable to have a provocation method with high specificity that avoids the use of pharmacologic provocations in large numbers of healthy children. It was the aim of this study to establish a range of values for responsiveness to cold-air challenge in normal children, to describe its specificity and sensitivity for a diagnosis of asthma in a large unselected sample of schoolchildren in a field study, and to measure factors associated with hyperresponsiveness. All fourth-grade schoolchildren (9 to 11 yr of age) in Munich and several southern Bavarian communities (n = 9,403) were surveyed with a questionnaire (response rate, 87%), baseline lung function, and cold-air hyperventilation challenge (n = 5,697). A diagnosis of asthma was reported in 7.9% of children. In the reference group, the ninety-fifth percentile (%ile) for the change of FEV1 (DFEV1) was -9.0%; the 90%ile was -7.3%. Asymptomatic children fulfilling all criteria for the reference group showed a significantly increased reactivity if they had a family history of asthma (p = 0.038). With a cutoff for DFEV1 of -9.0% or -7.3% the respective sensitivities to "detect" a diagnosis of asthma were 21.6 and 30.7%, and for asthma with symptoms during the previous 12 months it was 26.2 and 35.1%. Reactivity to cold air increased with the number of episodes of asthma in the previous 12 months. In a multiple linear regression model, factors associated with increased reactivity to cold air were diagnosed asthma, hay fever, male sex, younger age, and a positive skin test.(ABSTRACT TRUNCATED AT 250 WORDS)