MRI methods are widely used to follow the pathological evolution of multiple sclerosis in life and its modification by treatment. To date, measures of the number and volume of macroscopically visible lesions have been studied most often. These MRI outcomes have demonstrated clear treatment effects but without a commensurate clinical benefit, suggesting that there are other aspects of multiple sclerosis pathology that warrant investigation. In this context, there has been considerable interest in measuring tissue loss (atrophy) as a more global marker of the adverse outcome of multiple sclerosis pathology, whether it arises in macroscopic lesions or in the normal appearing tissues. An International Workshop recently considered the measurement of atrophy in multiple sclerosis and provided the basis for this review. Brain white matter bulk consists predominantly of axons (46%) followed by myelin (24%), and progressive atrophy implies loss of these structures, especially axons, although variable effects on tissue volumes may also arise from glial cell proliferation or loss, gliosis, inflammation and oedema. Significant correlations found between brain volume and other putative MR neuronal markers also indicate that atrophy reflects axonal loss. Numerous methods are available for the measurement of global and regional brain volumes and upper cervical cord cross-sectional area that are highly reproducible and sensitive to changes within 6-12 months. In general, 3D-T(1)-weighted acquisitions and largely automated segmentation approaches are optimal. Whereas normalized volumes are desirable for cross-sectional studies, absolute volume measures are adequate for serial investigation. Atrophy is seen at all clinical stages of multiple sclerosis, developing gradually following the appearance of inflammatory lesions. This probably reflects both inflammation-induced axonal loss followed by Wallerian degeneration and post-inflammatory neurodegeneration that may be partly due to failure of remyelination. One component of atrophy appears to be independent of focal lesions. Existing immunomodulatory therapies have had limited effects on progressive atrophy, concordant with their modest effects on progressive disability. Atrophy provides a sensitive measure of the neurodegenerative component of multiple sclerosis and should be measured in trials evaluating potential anti-inflammatory, remyelinating or neuroprotective therapies.