INTRODUCTION: The present study compared measurements of cardiac output by an arterial pressure-based cardiac output (APCO) analysis method with measurement by intermittent thermodilution cardiac output (ICO) via pulmonary artery catheter in a clinical setting. METHODS: The multicenter, prospective clinical investigation enrolled patients with a clinical indication for cardiac output monitoring requiring pulmonary artery and radial artery catheters at two hospitals in the United States, one hospital in France, and one hospital in Belgium. In 84 patients (69 surgical patients), the cardiac output was measured by analysis of the arterial pulse using APCO and was measured via pulmonary artery catheter by ICO; to establish a reference comparison, the cardiac output was measured by continuous cardiac output (CCO). Data were collected continuously by the APCO and CCO technologies, and at least every 4 hours by ICO. No clinical interventions were made as part of the study. RESULTS: For APCO compared with ICO, the bias was 0.20 l/min, the precision was 1.28 l/min, and the limits of agreement were -2.36 l/m to 2.75 l/m. For CCO compared with ICO, the bias was 0.66 l/min, the precision was 1.05 l/min, and the limits of agreement were -1.43 l/m to 2.76 l/m. The ability of APCO and CCO to assess changes in cardiac output was compared with that of ICO. In 96% of comparisons, APCO tracked the change in cardiac output in the same direction as ICO. The magnitude of change was comparable 59% of the time. For CCO, 95% of comparisons were in the same direction, with 58% of those changes being of similar magnitude. CONCLUSION: In critically ill patients in the intensive care unit, continuous measurement of cardiac output using either APCO or CCO is comparable with ICO. Further study in more homogeneous populations may refine specific situations where APCO reliability is strongest.