We have developed a rebreathing technique for measuring cardiac output in resting or exercising subjects. The data needed are the subject's CO2 dissociation curve, the initial volume and CO2 fraction of the rebreathing bag, and a record of CO2 at the mouth during the maneuver. From these one can obtain all the values required to solve the Fick equation. The combined error due to inaccuracy in reading the tracings and to the simplifying assumptions was found to be small (mean = 0.5%, SD ;.5%). Cardiac output values determined with this technique in normal subjects were on the average 2% higher than those obtained simultaneously with an acetylene rebreathing method (n = 49, SD = 11%). Among the advantages of the technique are that it requires analysis of a single gas, takes less than thirty seconds per determination, allows one to obtain repeated measurements at rapid intervals, is not affected by the ability of lung tissue to store CO2, and eliminates many of the assumptions usually made in non-invasive measurements of cardiac output.