We studied 143 young subjects by skin-prick testing to common allergens and by the measurement of non-specific bronchial responsiveness (NSBR). A logistic regression model showed a prevalence odds ratio (POR) for bronchial hyper-responsiveness (PD20 less than 10 mumol methacholine) in house dust positive subjects of 4.10 (95% CI 1.77-9.51, P less than 0.001), and in mould positive subjects 5.72 (95% CI 2.06-15.9, P less than 0.001): the PORs for epithelia (2.05, P = 0.12) and grasses (1.78, P = 0.19) were not significant. If bronchial reactivity was assessed as measurable (PD20 less than 102 mumol methacholine) or not, the POR for house-dust-positive subjects was 4.83 (95% CI 2.23-10.5, P less than 0.001), for moulds was 10.5 (95% CI 2.33-47.5, P less than 0.001), for epithelia was 4.79 (95% CI 1.91-12.0, P less than 0.001), and for grasses was 2.21 (95% CI 1.11-4.4, P = 0.022). The results show the risk of bronchial hyper-responsiveness is greater in subjects sensitive to house dust and mould than in those reactive to grasses, and suggests that the presence or absence of increased NSBR in atopic individuals may depend on the antigens to which they become sensitized.