The pathogenesis of ascites formation in cirrhosis is uncertain. It is still under debate whether the effective blood volume is reduced (underfilling theory) or whether the intravascular compartment is expanded (overflow theory). This problem has not yet been solved because of insufficient tools for measuring the central blood volume. We have developed a method that enables us to determine directly the central blood volume, i.e., the blood volume in the heart cavities, lungs, and central arterial tree. In 60 patients with cirrhosis and 16 control subjects the central blood volume was assessed according to the kinetic theory as the product of cardiac output and mean transit time of the central vascular bed. Central blood volume was significantly smaller in patients with cirrhosis than in controls (mean 21 vs. 27 ml/kg estimated ideal body weight, p less than 0.001; 25% vs. 33% of the total blood volume, p less than 0.0001). The lowest values (18 ml/kg) were found in patients with gross ascites and a reduced systemic vascular resistance. In patients with cirrhosis central blood volume was inversely correlated to the hepatic venous pressure gradient (r = -0.41, p less than 0.01), and the total blood volume was inversely correlated to the systemic vascular resistance (r = -0.49, p less than 0.001), the latter being significantly reduced in the patient group. Patients with cirrhosis apparently are unable to maintain a normal central blood volume. This may be due to arteriolar vasodilation, portosystemic collateral flow, or sequestration of fluid in the peritoneal cavity, or any combination thereof. The present results indicate that central circulatory underfilling is an integral part of the hemodynamic and homeostatic derangement observed in cirrhosis.