There was no correlation of blood volume measurements with central venous pressure (CVP) or hematocrit determinations and only minimal suggestive trends with wedge pressure in a large series of postoperative patients; the lack of correlations emphasize the unreliability of venous pressure and hematocrit determinations to predict blood volume alterations. To evaluate the physiological problems, to define optimal therapeutic goals, and to measure therapeutic effectiveness of volume loading with an oncotically active agent, we measured the hemodynamic and oxygen transport responses to 500 ml. of 5 percent albumin given over 1 hour in 22 patients with CVP greater than 15 cm. H2O. The patients were separated into two groups according to the CVP response to volume therapy. The CVP decreased in 14 (64 percent) of these patients (Group 1), but it increased slightly but not significantly in eight (36 percent) patients (Group 2). In Group 1 patients, there was increased flow, improvement of tissue perfusion as reflected by increased oxygen consumption, and augmentation of the ventricular function. In Group 2 there were slight increases in mean flow, mean pulmonary arterial pressure, and mean transit time and slightly decreased pulmonary vascular resistance; there was appreciable improvement in left ventricular function without significant deterioration of right ventricular function. The high initial central venous pressure is not a reliable index of either hypervolemia or cardiac failure in critically ill patients. It is concluded that a trial of volume loading with an oncotically active agent with frequent auscultation of the chest and careful observation of the CVP trends will give the maximum diagnostic as well as therapeutic information.