BACKGROUND: A noninvasive method for estimating cardiac output was tested in dogs. The technique is based on a differential CO2 Fick equation applied during normal ventilation and 50 seconds of partial rebreathing using additional dead space. We compared the cardiac output measured by the CO2 rebreathing method vs. the thermodilution technique. METHODS: Seven mongrel-dogs (24.6 +/- 0.4 kg) were studied, anesthesia was induced and maintained with a pentobarbital 25 mg/kg IV bolus injection followed by an infusion of 5 mg/kg/h. Mechanical ventilation was accomplished with a Servo 900C ventilator with FiO2 0.6 to maintain normocarbia. A fiberoptic pulmonary artery catheter was introduced via an external jugular vein for continuous monitoring of the cardiac output by the thermodilution method. Also cardiac output was measured by using partial CO2 rebreathing method. A continuous infusion of 0.5% bupivacaine was started at 0.5 mg/kg/min via the venous infusion port of the pulmonary catheter. Bupivacaine was infused continuously until mean arterial pressure decreased to 60 mmHg or less for at least 5 seconds. RESULTS: The total measured cardiac outputs (n = 72) were distributed over the range of 1.03-7.72 L/min by thermodilution method and 1.6-7.3 L/min by CO2 rebreathing. The correlation coefficients between both cardiac outputs was 0.6, the mean difference was 0.27 +/- 0.81 L/min. CONCLUSIONS: The cardiac output measured by CO2 rebreathing method was well correlated with cardiac output by thermodilution method.