Because of the major difficulties in measuring clinical end points in multiple sclerosis (MS) treatment trials, there has been much enthusiasm for using magnetic resonance imaging (MRI) findings as an alternative outcome. To provide international consensus guidelines for the use of MRI in MS clinical trials, a task force of the US National MS Society was convened. The recommendations of the task force are presented in this review. Given the high sensitivity for detecting pathological activity in relapsing-remitting and secondary progressive MS, monthly T2-weighted and gadolinium-enhanced brain MRI is an excellent tool for short-term exploratory trials of new agents where it serves as the primary end point; in particular, failure to demonstrate a reduction in lesion activity avoids the time, cost, and risks of a larger clinical end point study. However, conventional MRI findings have a limited correlation with disability in established MS. The primary end point of a definitive trial should therefore be clinical, although serial MRI at 6- to 12-month intervals is a useful secondary end point in providing an index of pathological progression. In trials of patients presenting with clinically isolated syndromes suggestive of MS, MRI findings can be used in the entry criteria, and as a secondary outcome measure, but conversion to clinically definite MS should be the primary outcome. The pathological substrates of irreversible disability are demyelination and axonal loss. Putative magnetic resonance markers for these processes include decreased N-acetylaspartate on proton magnetic resonance spectroscopy, decreased magnetization transfer ratios, hypointensity on T1-weighted images, and loss of short T2 water fractions, some of which relate more closely to disability than conventional MRI findings. Further technical developments should lead to more accurate quantitation, greater pathological specificity, and stronger clinical correlations.