the actual presence of diffusion-limitation for oxygen. Thus, significant inequality of alveolar ventilationperfusion ratios (V'A/Q') could equally explain a marked improvement in arterial Pa,O 2 with 100% oxygen breathing. Indeed, recent studies using the multiple inert gas elimination technique (MIGET), an analysis which can indirectly detect oxygen diffusion-limitation, have attributed the arterial hypoxaemia in patients with liver cirrhosis to V'A/Q' inequality and intrapulmonary shunting [9][10][11]. However, none of these MIGET studies has providedany direct evidence of actual pulmonary vascular dilatation; and, hence, they do not, by themselves, invalidate the specific hypothesis that diffusionlimitation for oxygen can occur as a direct consequence of abnormal dilatation of interalveolar vessels.To adequately test the above hypothesis requires an experimental approach that has the power to simultaneously Eur Respir J, 1995, 8, 2015-2021 DOI: 10.1183 ABSTRACT: To test the hypothesis that diffusion-limitation for oxygen is due to abnormal vascular dilatation and significantly contributes to the arterial hypoxaemia of liver cirrhosis requires an experimental approach that detects both diffusion-limitation for oxygen and the presence of abnormal dilatation of pulmonary vessels exposed to alveolar gas. We therefore studied the gas exchange of a 64 year old man with alcoholic liver cirrhosis and severe resting arterial hypoxaemia (arterial oxygen tension (Pa,O 2 ) 7.5 kPa) whilst breathing air and 100% O 2 using conventional blood gas (CBG) analysis, the multiple inert gas elimination technique (MIGET) and whole body scintigraphy (WBS) following the i.v. administration of radiolabelled boli of macroaggregates with a minimum diameter of 15 µM.During air breathing, there was a consistently positive difference between the arterial oxygen tension predicted by MIGET and that actually measured (P-M Pa,O 2 , average 0.9 kPa). During O 2 breathing, P-M Pa,O 2 became negative, (average -12.2 kPa), and shunt estimated by the O 2 method (% of Q') was consistently less than that measured by MIGET. Whereas both O 2 method and MIGET estimates of shunt never exceeded 25%, the WBS shunt was 40%, indicating that a substantial fraction of cardiac output flowed through abnormally dilated pulmonary vessels, some of which were exposed to alveolar gas and, hence, participated in gas exchange.Although our observations pertain to one subject, we believe they provide the most convincing in vivo evidence to date that abnormal dilatation of interalveolar vessels may, per se, result in a significant diffusion impairment for O 2 . Furthermore, in view of the consistently negative P-M Pa,O 2 observed during oxygen breathing, we speculate that such abnormal vascular dilatation may also have produced a significant diffusive impairment of one or more of the less soluble inert gases used in the MIGET analysis. Eur Respir J., 1995Respir J., , 8, 2015Respir J., -2021