Multiple sclerosis is an autoimmune disease characterized by demyelination and axonal loss. Conventional magnetic resonance imaging allows the demonstration of spatial and temporal dissemination of multiple sclerosis lesions earlier than is possible from clinical assessments. A variety of conventional magnetic resonance imaging protocols, in conjunction with clinical assessment, are now routinely used to increase the accuracy of diagnosis and long-term prognosis of multiple sclerosis. T2-weighted hyperintense lesions are related primarily to increased water content and thus cannot distinguish between inflammation, edema, demyelination, Wallerian degeneration, and axonal loss, whereas the contrast gadolinium-enhanced lesions on T1-weighted images reflect increased blood-brain barrier permeability associated with active inflammatory activity. Conventional magnetic resonance imaging metrics are not sufficiently sensitive to detect invisible brain damage in the normal appearing brain tissue, and they do not show a reliable correlation with clinical measures of disability. However, numerous studies showed that they can improve accuracy in the diagnosis and prognosis of multiple sclerosis. Recently, non-conventional magnetic resonance imaging techniques have been introduced to increase the accuracy of diagnosis and prognosis of multiple sclerosis. Several studies have used brain atrophy, T1-hypointense lesion volume, magnetization transfer imaging, diffusion-weighted imaging and magnetic resonance spectroscopy to test whether the extent and severity of tissue loss in lesions and in normal appearing gray and white matter at the time of clinically isolated syndrome may have diagnostic and prognostic value. These magnetic resonance imaging techniques represent a powerful tool to non-invasively study different pathological substrates of lesions and microscopic tissue changes. Additional short- and long-term prospective studies are requested to establish their value in the diagnosis and prognosis of multiple sclerosis.