Transpulmonary thermodilution (TPTD) using the injection of ice-cold saline is a reliable technique for measuring cardiac output in adults and children. We describe two cases in which these measurements were erroneously influenced by diffusion of the temperature indicator. If the thermistor-tipped arterial catheter is in close proximity to the central venous catheter, the injection of an icecold indicator might lead to an early temperature drop in the arterial catheter thus producing a biphasic dilution curve. This is called the "cross-talk phenomenon". In the first case, concerning a child undergoing congenital cardiac surgery, we show that this phenomenon is dependent on blood flow. In the second case we demonstrate an animal experiment in which the cross-talk phenomenon occurred during a very low flow state. However in this case the venous catheter was positioned in the inferior caval vein and not in close proximity to the arterial catheter. The cross-talk phenomenon can also occur when the arterial catheter is not in close position to the central venous catheter. We conclude that when using TPTD measurements with arterial and venous catheters both positioned in the lower body, the cross-talk phenomenon may occur. Therefore the thermodilution curve should always be closely monitored for a biphasic shape, especially when the cardiac output is low. Copyright © 2009, Nederlandse Vereniging voor Intensive Care. All Rights Reserved.