Many T cell surface proteins are attached to the cell membrane by glycosylphosphatidylinositol (GPI) anchors, which may be involved in cell signaling, protein targeting or protein release. Proteins destined to be GPI-anchored have both N- and C-terminal signal peptides, which permit the nascent polypeptides to transverse the endoplasmic reticulum and to be transferred to preformed GPI anchors in a transamidase reaction. The biosynthetic pathway of the GPI anchor in mammalian cells has been elucidated using a panel of T cell mutants that cannot express GPI-anchored proteins on their cell surfaces. The first step in anchor biosynthesis is the transfer of N-acetylglucosamine (GlcNAc) to a phosphoinositol (PI) acceptor to form GlcNAc-PI, which is then deacetylated to form GlcN-PI. Subsequently, a fatty acid is added to the inositol ring and three mannose residues are transferred to the elongating GPI core. Finally, ethanolamine phosphates are transferred to the mannose residues. Three human cDNAs encoding for GPI synthases (Classes A, F and H) have been identified using expression cloning technique. Mutation in the X-linked class A gene was shown to be the cause of paroxysmal nocturnal hemoglobinuria, an acquired disease affecting hematopoietic stem cells in which the abnormal cells are deficient in GlcNAc-PI formation. The potential involvement of GPI-anchored proteins or GPIs in T cell activation is also discussed.