Oligodendrocyte and astrocyte lineages were traced in rat forebrain sections using single- and double-label immunoperoxidase and indirect immunofluorescent techniques. Antibodies were directed against antigenic markers, the expressions of which overlapped in time: GD3 ganglioside in immature neuroectodermal cells; vimentin in radial glia; glial fibrillary acidic protein (GFAP) in astrocytes; and carbonic anhydrase (CA) and galactocerebroside (GC) in oligodendrocytes. A histochemical stain for iron was also used as a marker of oligodendrocytes. Small cells of the subventricular zone (SVZ) were stained with anti-GD3 but not with the other antibodies. By 16 d of gestation (E16), the SVZ generated large, round cells and thick, process-bearing cells that were GD3+/CA+/iron+. These cells then appeared in the cingulum and, with time, increased in numbers and extended thick processes as they filled the subcortical white matter. These cells eventually lost their reactivity to anti-GD3 but became GC+/CA+ with processes extending to myelin sheaths. At E15 radial glia were stained with the anti-vimentin antibody but were negative for GFAP. At birth, only the vimentin+ radial glia midline between the 2 ventricles were GFAP+, but with time more vimentin+ cells became GFAP+. By 7 d of postnatal age all the vimentin+ cells were GFAP+ and had converged predominately on the cingulum. With time these cells condensed and took on characteristic shapes of astrocytes. The embryonic separation of the oligodendrocyte and the astrocyte lineage is supported by four pieces of evidence: (1) GD3+ cells were double labeled with anti-CA, and then went on to become GC+; (2) vimentin+ and GFAP+ cells were not also GD3+; (3) ultrastructural localization of anti-GD3 was confined to cells with characteristics consistent with developing oligodendrocytes; and (4) the shapes of GD3+, CA+, GC+, or iron+ cells did not resemble those of the vimentin+ or GFAP+ cells.