Two previous uncontrolled studies have suggested that noninvasive mechanical ventilation (NIMV) in patients with hypercapnic chronic obstructive pulmonary disease (COPD) improves arterial blood gas tensions by decreasing lung hyperinflation with the consequent reduction in inspiratory loads and changes in ventilatory pattern. The aim of this randomised placebo-controlled study was to determine whether these mechanisms play a pivotal role in the effects of NIMV on arterial blood gases. Thirty-six stable hypercapnic COPD patients were randomly allocated to NIMV or sham NIMV. A 2-week run-in period was followed by a 3-week study period, during which ventilation was applied 3 h x day(-1), 5 days a week. Arterial blood gases, spirometry, lung volumes, and respiratory mechanics were measured before and after application of NIMV. Patients submitted to NIMV showed changes (mean (95% confidence interval)) in daytime arterial carbon dioxide tension (Pa,CO2) and arterial oxygen tension of -1.12 (-1.52-0.73) kPa (-8.4 (-11.4-5.5) mmHg) and 1.14 (0.70-1.50) kPa (8.6 (5.3-11.9) mmHg), respectively. Total lung capacity, functional residual capacity (FRC) and residual volume were found to be reduced by 10 (7-13), 25 (18-31), and 36 (27-45)% of their predicted value, respectively, whereas forced expiratory volume in one second and forced vital capacity increased by 4 (1.5-6.9) and 9 (5-13)% pred, respectively. Tidal volume (VT) increased by 181 (110-252) mL. All of the above changes were significant compared with sham NIMV. Changes in Pa,CO2 were significantly related to changes in dynamic intrinsic positive end-expiratory pressure, inspiratory lung impedance, VT and FRC. It was concluded that the beneficial effects of noninvasive mechanical ventilation could be explained by a reduction in lung hyperinflation and inspiratory loads.