Techniques for the measurement of cardiac output from soluble gas uptake by the lungs include the rebreathing method using nitrous oxide. The accuracy of this technique is well accepted, but its repeatability of measurement (precision) has not been well documented. We assessed the repeatability of measurements of pulmonary blood flow by the Innocor, a device employing the nitrous oxide rebreathing method. Successive paired measurements of pulmonary blood flow were made separated by a 5 min interval by the nitrous oxide rebreathing method, in 8 patients pre- or post cardiac surgery, and in 8 healthy volunteers. The standard deviation of the difference between first and second measurements was 0.84 l/min in the cardiac surgery group, and 1.25 l/min in the healthy volunteers. There was no significant bias in successive paired measurements of pulmonary blood flow in either the cardiac surgery patients (mean [95%CI] = -0.02 l/min [-0.62 to 0.57] or the healthy volunteers (0.00 l/min [-0.88 to 0.88]). Intra-class correlation coefficients for the healthy and cardiac patients were 0.77 and 0.64 respectively. Multiple measurements should be made and averaged when using the inert gas rebreathing technique for pulmonary blood flow determination. When comparing agreement with other methods for cardiac output measurement, the internal consistency of both methods should be considered.