Cryostat sections of two old plaques removed at autopsy from the spinal cord of a 62-year-old man with multiple sclerosis of 24-year duration were studied by indirect immunofluorescence with antibodies to neurofilament proteins, glial fibrillary acidic protein (GFAP), glial hyaluronate-binding protein (GHAP), vimentin and laminin. The neurofilament monoclonal antibodies used in this study reacted with phosphorylated epitopes of the two large polypeptides of the neurofilament triplet (NF 150K, NF 200K). As previously reported [Dahl D, Labkovsky B, Bignami A (1989) Brain Res Bull 22:225–232], the neurofilament antibodies either stained axons in the distal stump of transected sciatic nerve in the early stages of regeneration or late in the process, i.e., after regenerating axons had reached the distal stump of the transected sciatic nerve. Both multiple sclerosis plaques were positive for GFAP and vimentin, but negative for GHAP, while astrocytes in myelinated spinal cord white matter stained with both GFAP and GHAP antibodies. Laminin immunoreactivity in the plaques and normal spinal cord was confined to blood vessels. One plaque was almost devoid of axons as evidenced by indirect immunofluorescence with neurofilament antibodies. Another plaque was packed with bundles of thin axons running an irregular course in the densely gliosed tissue. Axons in the plaque only stained with neurofilament antibodies reacting with sciatic nerve in the early stages of regeneration while axons in the surrounding myelinated white matter were decorated by all neurofilament antibodies, regardless of the time of appearance of immunoreactivity in crushed sciatic nerve. It is concluded that reactive astrocytes forming glial scars do not constitute a non-permissible substrate for axonal growth.