BACKGROUND--In hospital clinics exercise challenge is used as a simple, non-invasive, non-pharmacological test for asthma in childhood. Little is known of its value in a community setting. An exercise test was therefore evaluated as an adjunct to a respiratory questionnaire in the course of an asthma survey. METHODS--From a cohort of 4003 primary school children, 607 of 799 with respiratory symptoms answered a detailed respiratory questionnaire. From the same cohort 135 of 229 randomly selected asymptomatic children were also interviewed. A stratified selection of one in four of the children interviewed was then invited to take part in a six minute cold air enhanced exercise challenge test; 128 symptomatic and 26 asymptomatic children attended. RESULTS--Bronchial hyperreactivity, a fall of FEV1 > or = 10% at five, 10, or 15 minutes following the exercise challenge, was identified in 15 of 128 symptomatic children and in one of 26 asymptomatic children. Bronchial hyperreactivity was found in only one of three children with frequent shortness of breath and one of five with frequent wheeze. It was found in 13 of 58 children whose parents were aware of the diagnosis of asthma; in 10 of 26 children who were on regular prophylactic treatment; in only 11 of 70 children with a history of exercise induced symptoms; and seldom in children with mild symptoms. Gestational age and ventilator support in the neonatal period were significant predictors of bronchial hyperreactivity. CONCLUSIONS--Exercise testing enhanced by cold air adds very little to a well designed respiratory questionnaire in community studies of asthma in childhood.