Background: Pulmonary oedema is life threatening. A method is needed for reliable measurement of lung water in man in order to clarify aetiology, pathophysiology, prognosis, and to evaluate treatments against pulmonary oedema.Methods: The double‐indicator dilution method with bolus injection and optical detection of heavy water and indocyanine green in arterial blood for the measurement of cardiac output, central blood volume and lung water was applied in two human and two experimental studies. In parallel, the thermo‐dye technique was tested in one of the experimental studies. The results from these four studies, together with results presented in the literature, were analysed according to the criteria set forth by The National Institute of Health, USA 1985, for the measurement of lung water in man, that is versatility, reproducibility, sensitivity, and accuracy.Results: for the heavy water‐indocyanine green method: Versatility: No adverse reaction was recorded in the two human studies. A central venous line and a peripheral arterial line, or access to an arterio‐venous fistula, were needed for bolus injection and concentration analysis. In‐line optical detection of heavy water and indocyanine green did not expose the subjects to radio‐labelled isotopes and the blood loss was low compared to timed blood volume collection. Reproducibility: The coefficient of variation was below 10% for cardiac output and central blood volume, and 10–17% for lung water. Sensitivity: Changes in lung water that were below the level of gas exchange disturbance could be detected in a group of subjects. Accuracy: Cardiac output, mean transit time and lung water for heavy water were linearly related to independent reference measures. Evidence for barrier limitations for heavy water in the clinical setting were weak. The indicator dilution method is unreliable for shock states and pulmonary embolism.Results: for the thermo‐dye method: Versatility: The method did not expose the subjects to radio‐labelled isotopes and there was no blood loss, but the method requires a central venous line and a thermistor‐equipped fibre‐optic catheter placed in the aorta.Reproducibility: The coefficient of variation was approximately 10% for cardiac output, central blood volume and lung water. Sensitivity: In healthy lungs, small differences in lung water can be detected. In pulmonary oedema, the large increase in deviation cripples the sensitivity of the method. Accuracy: Recovery of the thermal indicator and the error in its mean transit time were dependent on the distribution volume for the indicator cold. This undermines reliable calculation of cardiac output and lung water.Conclusions: The method using optical detection of heavy water and indocyanine green in blood for the measurement of cardiac output, central blood volume and lung water in humans is an improvement with regard to radio‐labelled methods, with the same reproducibility. The method allows for detection of small changes in lung water under the range where gas‐exchange disturbance occurs. The method provides values close to gravimetric lung water values in both healthy and grossly oedematous lungs, independent of cardiac output and distribution volume, with the exception of shock states and severe vascular obstruction.