Current concepts of the pathophysiology of ascites formation in cirrhosis of the liver have become more complex. Traditionally, the initiating event of renal sodium and water retention in cirrhosis was considered to be ascites formation ("underfilling" hypothesis) or primary renal dysfunction ("overflow" hypothesis). Changes in systemic, splanchnic and renal hemodynamics, as well as of volume regulating hormones observed in cirrhosis are compatible with a decrease in effective blood volume as suggested by the "underfilling" hypothesis. These changes, however, have been shown to precede ascites formation. This observation, together with the demonstration of an increase in total blood volume in cirrhosis prompted the "overflow" hypothesis. However, many studies are incompatible with this concept and, in addition, the agent causing primary renal sodium retention in cirrhosis still remains to be defined. The recently proposed "vasodilation" hypothesis reconciles the most salient features of both theories, proposing peripheral arterial vasodilation as the initiating event of decreased effective blood volume and renal sodium retention. Further studies are needed to elucidate the temporal relationship and more precisely define the character of hemodynamic, humoral and renal changes in cirrhosis of the liver.