This paper reviews recent data concerning the interactions among pulmonary edema, intrapulmonary shunt and cardiac output in acute hypoxemic respiratory failure. In canine oleic acid edema, a 5 mm Hg reduction in pulmonary wedge pressure significantly reduces edema, but a corresponding increase in colloid osmotic pressure does not. When pulmonary wedge pressure is lowered, cardiac output can be maintained with infusions of nitroprusside, dopamine or dobutamine. Each vasoactive agent improves ventricular pumping function, and the increase in cardiac output is due in part to peripheral circulatory actions of the drugs. Although pulmonary shunt increases with these vasoactive agents, increased shunt is due to their pulmonary vasoactivity but to the associated increase in pulmonary blood flow. Positive end-expiratory pressure reduces venous return by raising right atrial pressure, and it does not depress ventricular pumping function. Rather, positive end-expiratory pressure increases ventricular filling pressure t a given end-diastolic volume; it does not reduce and probably increases edema, yet it reduces shunt by redistributing the edema. These interpretations suggest several goals for cardiovascular management in acute hypoxemic respiratory failure: (1) the lowest pulmonary wedge pressure consistent with adequate cardiac output; and (2) the least positive end-expiratory pressure consistent with saturation of adequate circulating hemoglobin on nontoxic inspired oxygen.