Current standards for diagnosing multiple sclerosis (MS) are based on the 2018 revision of McDonald criteria. They rely on MRI detection (or clinical demonstration) of demyelinating lesions in the CNS, which are distributed in space (DIS) and in time (DIT). It is also a requirement that any possible known disease that produces demyelinating lesions is ruled out before applying McDonald's criteria.

This last requirement makes MS an ill-defined entity, whose borders change every time that a new disease is set apart. Some cases previously considered MS are now considered distinct conditions, like Neuromyelitis optica or antiMOG associated encephalomyelitis. Because of the requirement of distributed lesions, a single lesion (RIS) is not considered MS. For the same reason, the prodromal stage of MS (the unknown condition that causes the lesions) would not be considered as MS if it could be found.

Sometimes the diagnosis must be retrospective, relying on gradual worsening of neurological signs/symptoms, due to the lack of understanding of the pathogenicity driving disease progression. However, the only definite diagnosis of MS is post-mortem autopsy, where lesions typical of MS can be detected through histopathological techniques.

Multiple sclerosis is typically diagnosed based on the presenting signs and symptoms, in combination with supporting medical imaging and laboratory testing. It can be difficult to confirm, especially early on, since the signs and symptoms may be similar to those of other medical problems.

The McDonald criteria, which focus on clinical, laboratory, and radiologic evidence of lesions at different times and in different areas, is the most commonly used method of diagnosis. They came in in 2001, and have since been revised several times.

Clinical data alone may be sufficient for a diagnosis of MS if an individual has had separate episodes of neurologic symptoms characteristic of the disease. In those who seek medical attention after only one attack, other testing is needed for the diagnosis. The most commonly used diagnostic tools are neuroimaging, analysis of cerebrospinal fluid and evoked potentials. Magnetic resonance imaging of the brain and spine may show areas of demyelination (lesions or plaques). Gadolinium can be administered intravenously as a contrast agent to highlight active plaques and, by elimination, demonstrate the existence of historical lesions not associated with symptoms at the moment of the evaluation. Testing of cerebrospinal fluid obtained from a lumbar puncture can provide evidence of chronic inflammation in the central nervous system. The cerebrospinal fluid is tested for oligoclonal bands of IgG on electrophoresis, which are inflammation markers found in 75–85% of people with MS. The nervous system in MS may respond less actively to stimulation of the optic nerve and sensory nerves due to demyelination of such pathways. These brain responses can be examined using visual- and sensory-evoked potentials.

Previous diagnosis was by the Schumacher and Poser criteria. Some stated that the only definitive proof was an autopsy or biopsy where lesions typical of MS could be detected.

Studies have found increased interactions with health services by people who were diagnosed with MS several years later. The range of symptoms included pain and fatigue. Studies continue on whether MS has a prodromal or early phase, and could possibly be diagnosed and treated much earlier.