Oxygen therapy is frequently required for patients with acute lung injury. Paradoxically, administration of high concentrations of oxygen to such patients, while necessary, may itself result in hyperoxic lung injury. Therapeutic modalities are necessary which will increase the efficiency of oxygen uptake by the lung during periods of acute lung injury, in order to reduce the need for increased concentrations of administered oxygen. Inhalational agents which depend on ventilation for distribution might be of great benefit in reducing arterial hypoxemia occurring during such lung injury. These agents would be preferentially active in non-injured lung regions with intact ventilation. By causing vasodilation and augmented perfusion in well ventilated lung regions only, such agents would optimize matching of ventilation and perfusion and improve arterial oxygenation. Total systemic oxygen delivery would also be improved by reductions in PVR resulting in increased right and left ventricular function and improved cardiac output. Nitric oxide (NO), also called endothelial cell-derived relaxing factor, causes relaxation of arteriolar smooth muscle . Inhaled NO has been shown to reverse hypoxic vasoconstriction in normal animals. Inhaled NO, in the appropriate dose, enters smooth muscle surrounding arterioles and is metabolized as it induces myofibrillar relaxation. Any NO which does diffuse into the circulation is rapidly bound and inactivated by hemoglobin, preventing systemic vasodilation. Inhaled NO has the potential to improve pulmonary gas exchange during acute lung injury. We have developed a canine model of both lobar and diffuse bacterial pneumonia which closely simulates the acute lung injury and cardiopulmonary changes which occur in humans with pneumonia. We are now studying the cardiopulmonary effects of administration of inhaled NO in this model. Initial studies of nonlethal pneumonia showed that inhaled NO improved pulmonary hemodynamics but had no observable effect on gas exchange. Oxygenation abnormalities were mild in this model however, and additional studies are now being conducted in a lethal canine model of bacterial pneumonia.