The extent to which extravascular lung water (EVLW) is dependent on cardiac output was analysed in anaesthetized and mechanically ventilated pigs. EVLW was measured by thermal-dye dilution technique, by a fibreoptic thermistor catheter system (system 1), and by a thermistor catheter-external optical cuvette system (system 2). During baseline conditions, at which cardiac output was 3.65 l/min, and EVLW was 11.7 and 7.7 ml/kg b.w. with systems 1 and 2 respectively. A reduction of cardiac output to a mean of 1.90 l/min by the addition of halothane to the inspired gas did not significantly affect EVLW with system 1 (-5%) but increased EVLW by 39% (p less than 0.05) with system 2. An increase of cardiac output to a mean of 4.78 l/min by intravenous infusion of isoproterenol caused a small increase in EVLW with system 1 (14%; p less than 0.05) and a decrease with system 2 (10%; p less than 0.05). The dependence on cardiac output was the same whether the catheters were positioned centrally (aortic root) or peripherally (abdominal aorta). With system 1 the CO dependence was due to different time constants in thermistor and optical systems, and with appropriate phasing the dependence could be eliminated. With system 2 a large overestimation of the mean transit time difference between the two indicators was seen when cardiac output was low, resulting in overestimation of EVLW. It is concluded that the dependence of EVLW volume on cardiac output is an artefact due to technical problems in the design of the recording equipment rather than a reflection of pulmonary or vascular effects.