Cardiac output can be measured accurately by transpulmonary arterial thermodilution using the PiCCO (Pulsion Medical Systems, Munich, Germany) system with a femoral artery catheter. We have investigated the accuracy of a new 50 cm 4 French gauge radial artery catheter and the ability to use the system with a shorter radial catheter. We studied 18 patients who had undergone coronary artery surgery and made three simultaneous measurements of cardiac output by arterial thermodilution and with a pulmonary artery catheter. The radial catheter was withdrawn in 5 cm increments and the measurements were repeated. We found close agreement between arterial thermodilution and pulmonary artery thermodilution with a mean (SD) bias of 0.38 (0.77) l x min(-1). Arterial thermodilution became unreliable once the catheter had been withdrawn by more than 5 cm. We conclude that cardiac output measurement with arterial thermodilution with a radial catheter is interchangeable with that derived from a pulmonary artery catheter, and that a centrally sited arterial catheter is required for accurate determination of cardiac output by transpulmonary arterial thermodilution.