Purpose Although thermodilution cardiac output (CO_td) is the accepted method for monitoring patients with heart failure in the intensive care unit, it is particularly inaccurate in low-flow states and in the presence of tricuspid regurgitation; in addition, it does not adequately reflect the interaction between oxygen delivery and tissue oxygen extraction. We were concerned about the underuse of mixed venous oxygen saturation (SvO₂) in this setting and hypothesized that it might be a more useful parameter than CO_td for evaluating hemodynamic status and monitoring therapy in patients with severe congestive heart failure. Patients and Methods We determined the correlation coefficients between increasing dose of inotrope and each of the parameters CO_td and SvO₂ in 18 patients with congestive heart failure who received a 3-day trial of either milrinone or dobutamine during admission to the Veterans Affairs Medical Center in San Diego. In addition, we analyzed reproducibility of repeated CO_td and SvO₂ measurements taken at a single dose of inotrope. Results In patients receiving milrinone the increase in dose correlated much better with SvO₂ (average correlation coefficient 0.79) than with CO_td; this stronger correlation with SvO₂ was seen in 82% of the trials reviewed (p < 0.05). In addition, we found that repeated SvO₂ measurements taken at a single dose of milrinone were more reproducible than CO_td as indicated by smaller relative 95% confidence intervals. In patients receiving dobutamine no significant trend in correlation coefficients or reproducibility was observed. Conclusion The knowledge that there is a significant relation between SvO₂ and increasing dose of milrinone therapy in patients with severe heart failure and tricuspid regurgitation has practical value for physicians monitoring these patients in the intensive care unit. We believe this study demonstrates the worth of SvO₂ in detecting hemodynamic change during trials of milrinone therapy and that this parameter may in fact be more reproducible than traditional thermodilution methods. (Am Heart J 1998;135:383-8.)