Understanding the disease processes underlying multiple sclerosis is crucial to optimise treatment and to develop new therapeutic entities. Our understanding has been dominated by the inflammatory model of multiple sclerosis. More recently, a neurodegenerative model of the disease process has been developed which complements the inflammatory hypothesis in understanding the disease process and suggests a way forward to develop more effective treatments. Histopathological studies have shown that the early disease stage is characterised by acute inflammatory attacks, with T-cell infiltration, gliosis and acute demyelination. Axonal damage is also generally visible at this stage. In late-stage disease, continuing slow axonal damage may remain in the absence of signs of inflammation. Inflammation may not always have a deleterious outcome in multiple sclerosis since the release of growth factors from immune cells may protect neurones against axonal damage or facilitate axonal repair. The processes underlying lesion development appear to be heterogeneous, in some cases being driven by immune-cell mediated gliotoxicity and in others by primary gliopathy. Remyelination occurs to differing degrees in different individuals, and the reasons for this heterogeneity are poorly understood. Finally, the antigens that trigger the autoimmune response have not been characterised and may differ between patients. Candidates include myelin proteins, oligodendrocyte precursor proteins and axonal constituents. These different aspects of pathophysiology need to be brought together in a unified hypothesis of disease, but current knowledge does not permit such a hypothesis to be proposed.