Subdivision of the human liver into eight portal venous segments (according to Couinaud) is largely established in the anatomical and clinical community. However, this concept is challenged by an increasing number of surgical and radiological reports. We reexamined the intrahepatic portal venous architecture to understand the inconsistencies published. For this purpose, we studied the livers of 20 deceased who had donated their body to the Anatomy Department. The organs were investigated by portal venous injection, subsequent liver corrosion, and analysis of the branching pattern. After a usual bifurcation of the portal vein (order 0 vessel) into a right and left branch (first order vessels), the number of second order branches observed was between 9 and 44, with an average of 20. This seemingly trivial matter of fact suggests that the human liver does not consist of the eight segments presumed, but of many more. Supposedly contradictory observations turn out to be explainable by differing combinations of this large number of territories, and not simply by anatomical variability. For practical surgical purposes, we conclude that the useful eight-segment scheme needs conceptual reappraisal when a more realistic approach to the individual hepatic territoriality in the patients under consideration is demanded. We submit a "1-2-20-concept" as a possible key.