A new CO2-based non-invasive method for pulmonary blood flow has been developed. Selective rebreathing of CO2 was obtained in an open-circuit system by measuring the expired instantaneous CO2 flux in the expired air and by mixing pure CO2 into the inspired air. The time course of the inspired PCO2 was altered so that end-tidal PCO2 changed as a linear function of time (ramp). Pulmonary blood flow was computed as the ratio between the rate of change in net CO2 elimination (or uptake) and the concomitant rate of change of estimated arterial CO2 content. In comparison to simultaneously determined cardiac output by means of the O2 Fick method, the proposed CO2 ramp method underestimated cardiac output by some 20 per cent in 22 supine sedated patients with valvular heart disease. Differences in PCO2 between end-tidal gas and the gas in perfused alveoli are thought to be the main cause of this underestimation.