Background: Acute lung injury is characterized by hypoxemia which may be caused by hypoventilation, ventilation-perfusion ((V) over dot (A)/(Q) over dot) mismatch, intrapulmonary shunting and oxygen diffusion impairment. The multiple inert gas elimination technique (MIGET) allows analysis of these four causes of hypoxemia and is therefore the most comprehensive approach to investigate blood gas abnormalities. Using MIGET, we studied whether specific patterns of gas exchange abnormalities occur in different lung injury models and whether gas exchange abnormalities can be related to pathogenic aspects of lung injury. Methods: Lung injury was induced with oleic acid injection, endotoxin infusion or repeated lung lavage in groups of 6 mechanically ventilated pigs. Results: PaO2 decreased and PaCO2 increased significantly in all three lung injury models, but gas exchange was more impaired in the oleic add and lavage, as compared to the endotoxin group. Shunt was the major cause of hypoxemia in our lung injury models, whereas (V) over dot (A)/(Q) over dot mismatch contributed to venous admixture only after oleic acid injection and lung lavage. Oxygen diffusion limitation was not observed. Although alveolar ventilation was maintained after induction of lung injury, hypercapnia developed due to an increase of the ventilatory mean towards higher (V) over dot (A)/(Q) over dot ratios, increased shunt and increased carbon dioxide production. Conclusions: Shunt and ventilation-perfusion mismatch fully explain the gas exchange disturbances observed in our lung injury models. Although (V) over dot (A)/(Q) over dot distributions can be related to pathogenic aspects of the three study groups, we did not observe specific (V) over dot (A)/(Q) over dot patterns which allow diagnosis of the type of lung injury from a recovered (V) over dot (A)/(Q) over dot distribution.