To re-evaluate the accuracy and precision of a non-invasive method for measurement of cardiac output based on the differential CO(2) Fick approach using an automated change in respiratory rate delivered by a ventilator under control by a prototype measurement system.Twenty-four patients during coronary artery bypass surgery, pre- and postcardiopulmonary bypass were recruited. After routine cannulation including pulmonary artery catheter, relaxant general anesthesia was induced. After hemodynamic and ventilatory stability were achieved, simultaneous paired measurements were made by the differential Fick method and by bolus thermodilution. Measurements were generated by inducing a change in respiratory rate by the ventilator under computer control. In Group 1, this involved an increase in respiratory rate from 8 to 12 breaths/min. In Group 2, this involved a decrease from 12 to 6 breaths/min.Nineteen measurements were made in each Group, 12 pre-CPB and 7 post-CPB. In Group 1 mean bias was -0.06 l/min, with a precision of agreement of 0.87 l/min, r = 0.91. In Group 2 (excluding one outlier) mean bias was -0.07 l/min, with a precision of 1.12 l/min, r = 0.71.Acceptable agreement with thermo- dilution during surgery was found, particularly where the ventilatory change involved an increase in respiratory rate from a lower baseline. This approach has potential to be readily integrated into modern anesthesia delivery platforms, allowing routine non-invasive cardiac output measurement.