To evaluate the interaction between patient and ventilator during widely varying levels of pressure support (PS) ventilation, we studied 33 patients who had undergone aortocoronary bypass. All patients were without preoperative evidence of lung disease and had left ventricular ejection fractions greater than 45 percent. We assessed both changes in ventilatory pattern and the use of an extension of the Campbell technique to determine the components of the mechanical work of breathing (WOB). Patients were placed on 0, 10, 20, and 30 cm H2O of PS. We found that increasing the pressure support level (PSL) did not change minute ventilation, PCO2, or pH despite large changes in both rate and depth of breathing. The inspiratory time fraction was consistently and progressively reduced as PS increased. Although mean inspiratory flow (MIF) increased by 75 +/- 9 (SE) percent as the PSL increased to 30 cm H2O, mean airway pressure rose only 3.5 +/- 0.1 cm H2O. Observed changes in the resistive and elastic components of WOB at PSL greater than 0 were consistent with values predicted from baseline observations and changes in VT and MIF demonstrating that the Campbell technique of separating resistive and elastic components of the patient's WOB during unassisted ventilation can be extended to the analysis of WOB during mechanical ventilation. We were surprised to observe that although inspiratory WOB fell 67 +/- 13 percent as the PSL increased to 30 cm H2O, postinspiratory work by the inspiratory muscles (WOBPIIM) did not show significant change. The persistence and substantial values of WOBPIIM in some patients suggested the presence of significant patient-ventilator dyssynchrony, especially at higher levels of PS. Total inspiratory WOB per minute, including both patient WOB and WOB by the ventilator, increased by 186 +/- 29 percent, demonstrating that PS results in a respiratory pattern requiring substantially greater total mechanical work.