Proportional assist ventilation (PAV) has recently been proposed as a mode of synchronized partial ventilatory support. This study evaluates the short-term effects of nasal PAV on arterial blood gases in stable patients with chronic hypercapnia. Forty two patients (30 with chronic obstructive pulmonary disease (COPD) and 12 with restrictive chest wall disease (RCWD) due to kyphoscoliosis) underwent a 1 h run of nasal PAV. Randomly, two levels of assistance were performed: 1) PAV was set at a level corresponding to volume assist (VA) and flow assist (FA) at 80% of the individual values of elastance (Ers) and resistance (Rrs) obtained with the "runaway" method; and 2) VA and FA were set at a value corresponding to the difference between the patients' individual Ers and Rrs and normal values of Ers and Rrs. Arterial blood gases and dyspnoea (by visual analogue scale (VAS)) were evaluated in all patients during unsupported ventilation and 60 min of PAV. PAV was well tolerated and resulted in significant improvement in arterial oxygen tension (Pa,O2), arterial carbon dioxide tension (Pa,CO2) (6.8+/-0.8 to 7.4+/-1.4 and 7.2/-0.9 to 6.8+/-0.9 kPa, respectively) and VAS (29+/-23 to 20+/-18%). The effects of PAV were not different in the two groups of diseases nor in the two groups of settings. Different settings of nasal proportional assist ventilation are well tolerated and may improve gas exchange and dyspnoea in patients with stable hypercapnic respiratory insufficiency.