Indirect tests of bronchial responsiveness to agents such as adenosine 5'-monophosphate (AMP) or bradykinin might be more specific markers of a therapeutic responses to anti-inflammatory treatment than a test of direct responsiveness to agents such as methacholine. In children selected from the community on the basis of mildly symptomatic wheeze, we compared in a randomized, double-blind study design the effect of 400 micrograms/day of beclomethasone dipropionate (BDP) or placebo on three separate ways of provoking bronchial responsiveness, using methacholine, bradykinin, and AMP as the provoking agents. Following pretreatment bronchial challenges, 29 children received paired monthly methacholine and AMP challenges for 3 months, while for the same period another 33 children received paired monthly methacholine and bradykinin challenges. Compared with placebo-treated subjects, FEV1 increased significantly in the children receiving BDP. This improvement was observed in those randomized to either the AMP challenge or the bradykinin challenge. In children challenged with AMP, the PD20 AMP increased significantly after 1 month and 2 months of BDP therapy when compared with placebo, while under similar conditions the PD20 methacholine was not significantly affected. In children challenged with bradykinin, BDP therapy did not significantly alter either the PD20 bradykinin or PD20 methacholine. We conclude that a bronchial challenge with AMP appears to be a more sensitive predictor of response to anti-inflammatory treatment than either methacholine or bradykinin.