To our knowledge, no data have been provided as to whether and to what extent dynamic hyperinflation, through its deleterious effect on inspiratory muscle function, affects the perception of dyspnoea during induced bronchoconstriction in patients with chronic airflow obstruction. We hypothesized that dynamic hyperinflation accounts in part for the variability in dyspnoea during acute bronchoconstriction. We therefore studied 39 consecutive clinically stable patients whose pulmonary function data were as follows (% of predicted value): vital capacity (VC), 97.8% (S.D. 16.0%); functional residual capacity, 105.0% (18.8%); actual forced expiratory volume in 1s (FEV 1 )/VC ratio, 56.1% (6.3%). Perception of dyspnoea using the Borg scale was assessed during a methacholine-induced fall in FEV 1 . The clinical score and the treatment score, the level of bronchial hyper-responsiveness and the cytological sputum differential count were also assessed. In each patient, the percentage fall in FEV 1 and the concurrent Borg rating were linearly related, with the mean slope (PD slope) being 0.09 (0.06). The percentage fall in FEV 1 accounted for between 41% and 94% of the variation in the Borg score. At a 20% fall in FEV 1 , the decrease in inspiratory capacity (ΔIC) was 0.156 (0.050) litres. Patients were divided into three subgroups according to the PD slope (arbitrary units/% fall in FEV 1 ): subgroup I [eight hypoperceivers; PD slope 0.026 (0.005)], subgroup II [26 moderate perceivers; 0.090 (0.037)] and subgroup III [five hyperperceivers; 0.200 (0.044)]. By applying stepwise multiple regression analysis with the PD slope as the dependent variable, and other characteristics (demographic, clinical and functional characteristics, smoking history, level of bronchial hyper-responsiveness and sputum cytological profile) as independent variables, ΔIC ( r 2 = 45%, P < 0.00001) and to a lesser extent treatment score ( r 2 = 17.3%, P < 0.0006), and to an even lesser extent age ( r 2 = 3%, P < 0.05), independently predicted a substantial amount ( r 2 = 65.27%, P < 0.00001) of the variability in the Borg slope. Thus acute hyperinflation, and to a lesser extent treatment score and age, account in part for the variability in the perception of dyspnoea after accounting for changes in FEV 1 during bronchoconstriction in patients with chronic airflow obstruction.