Historically considered to be an autoimmune demyelinating disease, multiple sclerosis is now recognized to be characterized by significant axonal and neuronal pathology. Addressing this neurodegenerative component of the disease is an important treatment objective, since axonal injury is believed to underlie the accumulation of disability and disease progression. The precise relationship between the inflammatory and neurodegenerative components in multiple sclerosis remains poorly elucidated, although neurodegeneration appears to be at least partially independent from neuroinflammation. The mechanisms underlying axonal injury appear complex and are likely to be multifactorial. Specific treatment strategies need to be developed that act within the central nervous system to prevent neurodegeneration and need to be provided from the earliest stages of disease. It is likely that immunomodulatory treatments acting purely in the periphery will provide only indirect and not direct neuroprotection. A promising approach is to enhance neuroprotective autoimmunity inside the brain, believed to be mediated, at least in part, by the release of neurotrophic factors within the nervous system from infiltrating immune cells. Such a beneficial process would be inhibited by a non-selective immunosuppressive strategy. In summary, treatments of multiple sclerosis should take into account the heterogeneous pathophysiology of the disease. The pathogenic process in the central nervous system itself should be the major focus in multiple sclerosis therapy in order to protect against demyelination and axonal loss and to promote remyelination and regeneration directly in the target tissue, independently of peripheral immune status. In conclusion, selective treatment strategies aimed at preventing axonal injury within the central nervous system are required to complement existing, peripherally acting treatments targeting the immune system.