There is considerable debate on the development of a glial cell line in the rat optic nerve, which is characterized by the specific expression of the A2B5 and HNK-1 epitopes. This cell line has been assumed to give rise to oligodendrocytes and so-called type 2 astrocytes. However, it is doubtful that the latter cell type really exists in vivo. In the present study, we have addressed this question by investigating the development of astrocytes in the myelindeficient (md) rat, which is characterized by dysmyelination and loss of oligodendrocytes. Defective oligodendrocytes were observed by the third postnatal day, well before the generation of type 2 astrocytes. Consequently, the number of type 2 astrocytes was reduced in cultures prepared from optic nerves of md rats vs. controls. This finding was not paralleled in vivo; i.e., no dying astrocytes were observed in md sections by conventional electron microscopy. However, immunoreactivity against the HNK-1 epitope was enhanced in md compared to control sections. Ultrastructurally, HNK-1 immunoreactivity was detected predominantly on the axonal surface at astroaxonal contact sites, which were found only at the nodes of Ranvier within controls but extended to the whole axonal surface in md animals. Only a minor portion of the immunoreactivity derived from glial cells, presumably from oligodendrocytes at the paranodal region in controls. Thus, the HNK-1 epitope is not a useful antigen for distinguishing astrocytes in the rat optic nerve. Accordingly, our results do not provide evidence for the existence of specialized type 2 astrocytes in vivo. In vitro, these cells are probably only oligodendrocytes that mimic some astroglial features if grown in serumcontaining media.