The purpose of this review is to summarize recent studies that have led to a more complete understanding of the molecular basis of paroxysmal nocturnal hemoglobinuria (PNH). Somatic mutations of PIG-A arising in pluripotent hematopoietic stem cells are necessary for the development of PNH. PIG-A is an X-linked gene that is essential for synthesis of the glycosyl phosphatidylinositol (GPI) moiety that serves as a membrane anchor for a functionally diverse group of cell surface proteins. Consequently, the progeny of stem cells with mutant PIG-A are deficient in all GPI-anchored proteins (GPI-AP). Among the GPI-AP that are expressed on hematopoietic cells are two important regulators of the complement system, decay-accelerating factor, (CD55) and membrane inhibitor of reactive lysis, (CD59). It is the deficiency of erythrocyte CD55 and CD59 that accounts for the intravascular hemolysis and hemoglobinuria that are the clinical hallmarks of PNH. A remarkable feature of PNH is that the peripheral blood is a mosaic composed of variable proportions of GPI-AP+ and GPI-AP- cells and that, in an individual patient, the GPI-AP- cells can be derived from multiple mutant stem cells. Currently, however, there is no evidence that the PIG-A mutation per se provides a proliferative advantage. Thus, PNH is not a monoclonal disease with a malignant phenotype. Rather, the mutant stem cells appear to dominate hematopoiesis because under some pathological conditions, GPI-AP deficiency is advantageous. The close association of PNH with aplastic anemia suggests that the selection pressure arises as a consequence of a specific type of bone marrow injury.