The vascular effects of acetylcholine were evaluated in blood-free lungs of rabbits perfused in situ at a constant flow rate of 15 ml/min with Kreb''s-bicarbonate solution containing 3% dextran. Infusion of 1 .mu.M acetylcholine normally elicited vasoconstriction. However, acetylcholine (3.0 nM-1.0 .mu.M) produced concentration-dependent vasodilatation when administered to lungs in which pulmonary artery pressure (i.e., perfusion pressure) was elevated and cyclooxygenase activity was blocked, respectively, by infusing the endoperoxide analog U46619 (2.0-10.0 nM) and 30.0 .mu.M indomethacin. Three additional groups of lungs treated with indomethacin and preconstricted with U46619 were challenged twice with 1.0 .mu.M acetylcholine. Two of the groups were equilibrated with an infusion of either 20.0 .mu.M quinacrine or 10.0 .mu.M ferrous hemoglobin before and during their second challenges with acetylcholine and a control group received no treatment between trials. Neither quinacrine nor hemoglobin altered base-line pulmonary artery pressure, but both agents reversed the effect of acetylcholine from vasodilatation to vasoconstriction. This study agrees with previous reports of a pulmonary vasoconstrictor action of acetylcholine dependent on an intact cyclooxygenase pathway, but also provides new evidence for a vasodilator action of acetylcholine in the intact lungs of rabbits. Quinacrine and hemoglobin are known to antagonize the endothelium-dependent relaxation of vascular smooth muscle elicited by acetylcholine; therefore, this study provides indirect evidence supporting a role for the endothelium-derived relaxaing factor in the regulation of pulmonary vascular tone.