Responses to inhaled nitric oxide (iNO) in acute lung injury (ALI), as evidenced by improvements in oxygenation, are variable. We hypothesized that the effect of iNO may be related to the pre-iNO distribution of pulmonary blood flow (PBF). In the present study we evaluated the effect of iNO on PBF in normal healthy dogs and in a canine model of ALI induced by oleic acid (OA). In Group "OA only" (n = 5), ALI was induced by central venous injection of 0.08 ml/kg OA. In Group "E+OA" (n = 5), hypoxic pulmonary vasoconstriction after ALI was blocked with low-dose endotoxin (15 microg/kg of Escherichia coli endotoxin) administered 30 min before giving the same dose of OA. Measurements of regional PBF and lung water concentration (LWC) using positron emission tomography (PET) and H215O were performed before and after OA or placebo, and then again at concentrations of 10, 40, and 0 ppm iNO. One hundred twenty minutes after OA injury, PaO2/FIO2 fell significantly in Group OA only, from 567 +/- 32 to 437 +/- 67 mm Hg. In these animals, PBF redistributed from the dorsal edematous regions of the lungs to the nondependent zones, thus partially preserving normal ventilation/ perfusion relationships. As in the normal animals, in Group OA only, iNO did not significantly change either PBF or oxygenation. In Group E+OA, the administration of low-dose endotoxin eliminated perfusion redistribution from the dorsal edematous lung regions. As a result, PaO2/FIO2 fell from 558 +/- 70 to 119 +/- 53 mm Hg, a decrease that was significantly greater than that in Group OA only. In Group E+OA, administration of iNO restored perfusion redistribution to a similar level as in Group OA only, which was associated with a significant improvement in PaO2/FIO2, from 119 +/- 53 to 251 +/- 159 (10 ppm iNO), and 259 +/- 165 mm Hg (40 ppm iNO). We conclude that the effect of iNO on oxygenation after ALI depends on the pre-iNO perfusion pattern, which may help explain the variable response to iNO often observed in patients with acute respiratory distress syndrome.