There is no single test that is diagnostic of MS, including MRI. The lesions detected with MRI are pathologically nonspecific. The principles of MS diagnosis are based on showing dissemination of white matter lesions in space and time. MRI is the most sensitive method for revealing asymptomatic dissemination of lesions in space and time. The pattern and evolution of MRI lesions, in the appropriate clinical setting, has made MRI abnormalities invaluable criteria for the early diagnosis of MS. The first important role for MRI in the diagnosis of MS allows for an early diagnosis of MS for CIS patients using the IP diagnostic criteria, including MRI for dissemination in space (DIS) and time (DIT). The sensitivity of diagnosing MS within the first year after a single attack is 94%, with a specificity of 83%. The MRI evidence required to support the diagnosis varies, depending on the strength of the clinical findings. Allowing a new MRI lesion to substitute for a clinical attack doubles the number of CIS patients who can be diagnosed as having MS within 1 year of symptom onset. Increasing the sensitivity of the test with more lenient criteria, as recommended by the AAN subcommittee, can result in decreased specificity. The second important role for MRI in the diagnostic work-up of suspected MS patients is to rule out alternative diagnoses obvious on MRI, such as spinal stenosis and most brain tumors. Characteristic lesions that favor MS include Dawson Fingers, ovoid lesions, corpus callosum lesions, and asymptomatic spinal cord lesions. However, other white matter diseases can have similar appearances on MRI. Persistent gadolinium enhancement greater than three months, lesions with mass effect, and meningeal enhancement suggest other disorders. A standardized MRI protocol for brain and spinal cord is crucial for comparing across studies or between centers. T2W MRI cannot distinguish between acute and chronic lesions. Gadolinium provides useful information about new lesion activity and is helpful in ruling out alternative diagnoses such as neoplasm, vascular malformations, and leptomeningeal disease. A single gadolinium-enhanced MRI can potentially provide evidence for dissemination in space and time. Spinal cord imaging is equally valuable to rule out spinal stenosis or tumor, and for detecting asymptomatic lesions when brain imaging is nondiagnostic in patients suspected of having MS. Precise criteria may be too suggestive that MS can be diagnosed by MRI and a negative MRI at the time of CIS does not rule out MS. MRI evidence plays a supportive role in what is ultimately a clinical diagnosis of MS, in the appropriate clinical situation, and always at the exclusion of alternative diagnoses.