One hundred twenty-five separate cardiac output determinations were obtained after open-heart surgery in 10 patients by simultaneous use of thermodilution and dye-dilution techniques. Mean thermodilution cardiac output was 1.6 per cent greater than mean dye-dilution cardiac output (5.24 versus 5.16 L. per minute). Reproducibility of thermodilution cardiac output (coefficient of variation, 8.6 per cent) was superior to that of dye-dilution cardiac output (coefficient of variation, 12.3 per cent) for outputs ranging from 2.5 to 8.7 L. per minute (p less than 0.001). Linear regression analysis revealed a correlation showing that COtd = 0.86 COdye + 0.80 (r = 0.9, p less than 0.001) and indicating a similarity between thermodilution and dye-dilution output figures except in extremely low output states. In such cases, thermodilution cardiac output becomes progressively larger than dye-dilution cardiac output. The results indicate that thermodilution cardiac output is a valid method for determining cardiac output in the rapidly changing clinical setting following cardiopulmonary bypass. Clinical applications of this technique include evaluation of the efficacy of inotropic agents, effectiveness of intra-aortic balloon counterpulsation, and status of the low output syndrome postoperatively. Routine use in patients with Class III or IV cardiac disease appears justified.