OBJECTIVE To test the validity of using assumed oxygen consumption for Fick cardiac output during administration of epoprostenol. METHODS In 24 consecutive patients Fick cardiac output calculated with assumed oxygen consumption according to LaFarge and Miettinen (COLM) and according to Bergstra et al. (COBE) were compared with thermodilution cardiac output (COTH). Pulmonary vascular resistance (PVR) was calculated with each cardiac output (CO) value. If PVR exceeded 200 dyne.s.cm-5, administration of epoprostenol (Ep) was started, and at maximal dose the above-mentioned measurements were repeated. RESULTS In all 24 patients COBE agreed significantly with COTH, mean difference -0.145 1.min-1, 95% confidence interval (CI) -0.402 to 0.111, limits of agreement (LA) -1.336 to 1.045. COLM was significantly lower than COTH, -1.165 1.min-1, p<0.05, 95% CI -1.510 to -0.819, LA -2.768 to 0.438. In 16 patients (67%) administration of epoprostenol was indicated. During Ep infusion the CO values calculated with oxygen consumption according to LaFarge and Miettinen (EpCOLM) were also significantly lower than thermodilution CO (EpCOTH), mean difference -1.281 1.min-1, p<0.05, 95% CI -1.663 to -0.900, LA -2.685 to 0.122. The agreement of CO values calculated with oxygen consumption according to Bergstra et al. (EpCOBE) and EpCOTH remained, mean difference -0.115 1.min-1, 95% CI -0.408 to 0.178, LA -1.191 to 0.962. CONCLUSIONS Before as well as during administration of epoprostenol, it is justified to use CO values calculated with oxygen consumption according to Bergstra et al. instead of thermodilution CO; CO values calculated with oxygen consumption according to LaFarge and Miettinen show significant underestimation.