According to present concepts, the ascites that occurs in patients with cirrhosis of the liver is formed by transudation from the plasma into the peritoneal cavity through the walls of the capillaries of the portal system. A ready explanation for this transudation has been at hand, based upon the hypothesis originally proposed by Starling (1 to 3) that interchange of fluid between blood and tissue spaces is controlled by the balance between capillary blood pressure and osmotic pressure of plasma proteins. In patients with cirrhosis and ascites the portal venous pressure is high (4); the serum albumin concentration (5 to 7) and the serum colloid osmotic pressure (8) are low. The observations of Post and Patek (7) have suggested that the presence or absence of ascites in patients with cirrhosis of the liver is determined chiefly by the serum albumin level.Recent observations have thrown doubt upon this concept (9, 10). In the course of an investigation of the effects of intravenous injection of concentrated human serum albumin in patients with cirrhosis and ascites (10) it was observed that elevation of the colloid osmotic pressure of the plasma to normal levels for periods of 2 weeks did not decrease the rate of ascites formation. There occurred instead a marked increase in the rate of diffusion of albumin from the plasma into the ascitic fluid and a rise in the colloid osmotic pressure of the ascitic fluid which paralleled that of the plasma.Because these observations seemed inconsistent with the accepted concepts of the mechanism of ascites formation, the present studies were undertaken in order (a) to determine whether or not the formation of ascites in patients with cirrhosis of the liver is controlled by osmotic factors, in accordance with the Starling hypothesis, and (b) to evaluate the role of the diffusion of plasma proteins into the peritoneal cavity.According to the Starling hypothesis, the flow of fluid across the portal capillary membrane is controlled by the balance of hydrostatic and colloid osmotic pressures on both sides of the membrane (11), such that at equilibrium: portal capillary pressure minus intra-abdominal hydrostatic pressure = plasma colloid osmotic pressure minus ascitic colloid osmotic pressure.Disturbance of the osmotic balance by alteration of any of these factors should produce corresponding changes in the rate and direction of flow.In order to test the validity of this hypothesis, direct measurements were made of (1) the plasma colloid osmotic pressure, (2) ascitic colloid osmotic pressure, (3) the intra-abdominal hydrostatic pressure, and (4) the ascitic fluid volume, both in the unmodified course of the formation and loss of ascites, and after experimental alteration of one of the osmotic factors. The experimental elevation of plasma colloid osmotic pressure was effected by (a) mercupurin diuresis and (b) intravenous injection of concentrated human serum albumin. Plasma colloid osmotic pressure was decreased by the oral administration of physiological saline solution. Ascitic co...