How MS is diagnosed: MRI, lumbar puncture and the 2024 McDonald criteria

Short answer

MS is diagnosed by a neurologist, not by one test. You get an MRI of the brain and spinal cord, usually a lumbar puncture, and sometimes eye tests. The 2024 McDonald criteria turn those findings into a diagnosis: typical damage in at least two of five locations, with no better explanation (100270-4), 2).

If this has already taken longer than you expected, you're not imagining it and you're not being difficult. In a US patient registry of 428 people with MS, 40% said at least one of their early symptoms was first attributed to something else, and half said they had earlier symptoms that nobody evaluated at the time — a gap with a median of 5.4 years (11). Going slowly is also deliberate: MS is a diagnosis of exclusion, and the committee that writes the criteria says plainly that both misdiagnosis and underdiagnosis harm people (2).

Note: this article explains how clinicians reach an MS diagnosis and is not medical advice; a neurologist decides what your tests mean. Call 911 for sudden numbness or weakness of the face, arm or leg (especially on one side), sudden confusion or trouble speaking, sudden trouble seeing, sudden trouble walking or loss of balance, or a sudden severe headache with no known cause — these are stroke warning signs. Get emergency care the same day for new weakness in both legs, a band of numbness or tightness around your torso, or new trouble controlling your bladder or bowels. Get urgent care for new eye pain with loss of vision in one eye (15, 16, 17).

The MS workup at a glance

Here is the whole toolkit, and what each piece is actually for.

There is no blood test that says "you have MS." NINDS puts it directly: there is no single test for MS, and clinicians use several tests alongside the history and exam to rule the diagnosis in or out (7).

What does an MS MRI actually look for?

It looks for small areas of damage in specific places, not for "spots on the brain" in general. MRI is the backbone of the diagnosis: the 2024 criteria state that brain and spinal cord MRI remain the most useful paraclinical tests, and the accompanying imaging guidance says diagnostic imaging should always cover both the brain and the spinal cord (2, 400304-7)).

The five locations

MS lesions cluster in characteristic spots. The 2024 criteria recognise five:

• Periventricular — around the fluid-filled ventricles in the middle of the brain.

• Juxtacortical or cortical — at or in the outer layer of the brain.

• Infratentorial — the brainstem and cerebellum, at the back and base of the brain.

• Spinal cord.

• Optic nerve — new in 2024 (100270-4), 400304-7)).

Typical lesions in at least two of those five locations satisfy what the criteria call dissemination in space — and since 2024 it no longer matters whether the lesion is the one causing your symptoms (2).

Why they inject contrast

Gadolinium contrast highlights lesions where the blood–brain barrier is currently leaky, which usually means the lesion is new — roughly the last few weeks. A scan showing both an enhancing lesion and a non-enhancing one is showing damage of two different ages at a single sitting, which is how one MRI can satisfy dissemination in time (100270-4), 7).

Why the radiologist may want to rescan you

MS protocols are specific. Imaging guidance from MAGNIMS, the Consortium of Multiple Sclerosis Centers and NAIMS recommends fat-saturated sequences to pick up symptomatic optic nerve lesions, and susceptibility-sensitive sequences so the central vein sign and paramagnetic rim lesions can be assessed when conventional images are not enough (400304-7)). A scan ordered for headaches at an urgent care centre usually does not include any of that. "You already had an MRI" and "you already had the right MRI" are different sentences.

What white spots do not mean

Plenty of things make white matter lesions: migraine, high blood pressure, smoking, diabetes, normal ageing. That overlap is exactly why the 2024 criteria added extra safeguards for people aged 50 and over and for people with vascular risk factors or a headache disorder — in those cases the committee strongly recommends confirming the diagnosis with an additional feature: a spinal cord lesion, positive CSF, or a positive central vein sign (2).

What happens during an MS spinal tap, and do you need one?

A lumbar puncture takes CSF from the lower back, below where the spinal cord ends, to look for signs that your immune system is making antibodies inside your central nervous system. You may not need one — but it is frequently the thing that resolves an uncertain case.

What the procedure is like

You sit leaning forward or lie curled on your side. The skin over the lower spine is cleaned and numbed with local anaesthetic, a thin needle goes between two vertebrae into the fluid space, and a few millilitres of clear fluid drip out. The collection itself usually takes a few minutes. Blood is drawn at the same visit, because the CSF result only means something next to a matching blood sample (5).

The complication people ask about is the headache afterwards — a headache that comes on when you stand up and eases when you lie down. Needle design matters a lot here. A Cochrane review of 66 randomised trials and 17,067 participants found that cutting ("traumatic") needles roughly doubled the risk of post-dural puncture headache compared with atraumatic needles, with a risk ratio of 2.14, and atraumatic needles did not increase side effects such as backache or pins and needles (13). It is entirely reasonable to ask which needle your clinic uses.

What the lab is looking for

Two things, and under the 2024 criteria they are interchangeable:

Oligoclonal bands (OCBs). Your CSF and your blood are run side by side. If distinct bands of antibody appear in the CSF but not in the blood, antibodies are being made inside the nervous system rather than arriving from elsewhere. In a meta-analysis of 71 studies, 87.7% of 12,253 people with MS and 68.6% of 2,685 people with a clinically isolated syndrome were OCB-positive, and OCB-positive CIS carried almost ten times the odds of going on to MS (12).

Kappa free light chains (kFLC). Light chains are fragments that antibody-producing cells release in excess. The kFLC index compares the CSF-to-blood ratio of kappa light chains against the CSF-to-blood ratio of albumin, correcting for how leaky your blood–brain barrier happens to be. The 2024 criteria accept a kFLC index of 6.1 or above as equivalent to positive OCBs (2, 5).

This is the single biggest practical change to CSF testing in years. OCB testing is manual gel work; one laboratory director describes running only four CSF/serum pairs per gel. kFLC is automated and quantitative, and can turn around within 24 hours (19, 2). The two agree about 90% of the time, so in a discordant or atypical case your neurologist may want both (5).

What a negative result does not mean

A negative spinal tap does not clear you. Around one in eight people with MS never has oligoclonal bands (12). CSF is one input among several; it carries weight when it is positive and much less weight when it is not.

One more thing the fluid does: it screens for infections and other inflammatory conditions that mimic MS, which is part of why a tap is still offered even when the MRI already looks convincing (7).

What are the McDonald criteria, and what changed in 2024?

They are the international rulebook that decides whether your test results add up to MS, and the current edition is the 2024 revision, published in The Lancet Neurology on 18 September 2025 (100270-4), 3). If you have been reading older articles, they describe the 2017 version (930470-2)).

The criteria come from the International Advisory Committee on Clinical Trials in Multiple Sclerosis, a body sponsored by ECTRIMS and the National MS Society that has written every edition since 2001. The 2024 revision was agreed by 56 experts from 16 countries at a review meeting in Barcelona in 2023, using a voting process where each statement needed 80% agreement to be accepted (2). The National MS Society publishes a plain-language summary of the same criteria (8).

The changes, in plain language

• The optic nerve is now a fifth location for dissemination in space — A past bout of optic neuritis can count toward the diagnosis, shown by MRI, OCT or visual evoked potentials (100270-4), 600275-3))

• Dissemination in time is no longer mandatory in every case — In some situations you no longer have to wait for a second attack or a new lesion (2)

• Four of five locations can be enough on its own — With a typical presentation and typical lesions in at least four locations, no extra evidence is needed (2, 400304-7))

• kFLC index can replace oligoclonal bands — A faster, automated CSF test, cutoff 6.1 (5)

• Central vein sign and paramagnetic rim lesions can be used — Supportive MRI markers that raise specificity in ambiguous cases (100270-4))

• Radiologically isolated syndrome can be MS — If certain criteria are met, incidental MRI findings plus supporting evidence can be a diagnosis (2)

• One framework for everyone — Children and adults, relapsing and primary progressive, all use the same criteria (100270-4))

• Extra confirmation at 50+ or with vascular risk — A cord lesion, positive CSF or a positive central vein sign is strongly recommended (2)

The stated goal is a faster, easier, more accessible diagnosis without losing specificity (2). One caveat worth knowing: the committee itself lists the practical problems — the new MRI features need extra scanning and reporting time, and reading them takes expertise (2). A survey of 24 Portuguese centres in 2026 found central vein sign and paramagnetic rim lesions were not routinely analysed in 37.5% of them, and the kFLC index was routine in only 41.7% (22). Availability where you live is a real variable.

Dissemination in space and in time, in plain words

Two phrases run through every MS report, and they are simpler than they sound.

Dissemination in space means the damage is in more than one place. One lesion could be anything. Lesions in two or more of the five MS locations — periventricular, juxtacortical or cortical, infratentorial, spinal cord, optic nerve — is a pattern. Since 2024, it does not matter whether a given lesion is the one causing your symptoms; symptomatic and silent lesions both count (2).

Dissemination in time means the damage happened on more than one occasion. Classically this needed either a second clinical attack or a new lesion on a follow-up scan. It can also be shown on a single MRI, if that scan contains both an enhancing lesion and a non-enhancing one — damage of two ages, in one picture (100270-4)).

The 2024 edition is where this loosens. Routes to a diagnosis now include:

• dissemination in space and dissemination in time — as in 2017;

• dissemination in space and positive CSF (OCBs or kFLC index);

• typical presentation with typical lesions in at least four of the five locations, with nothing further needed;

• typical presentation, lesions in one region, plus a positive central vein sign or at least one paramagnetic rim lesion, plus either dissemination in time or positive CSF;

• in progressive disease, two spinal cord lesions count as dissemination in space, and primary progressive MS still requires at least 12 months of clinical progression (2).

And one condition sits above all of them: no better explanation. MS remains a diagnosis of exclusion (2).

Central vein sign and paramagnetic rim lesions: the newer MRI markers

These are two patterns visible on susceptibility-sensitive MRI that help tell MS lesions from look-alikes. Neither is required, and both are about specificity rather than catching more cases.

Central vein sign (CVS). MS lesions tend to form around a small vein, which shows up as a dark line through the middle of the lesion. The sign appears in roughly 65% of people with MS and in a much smaller share of lesions from migraine, small vessel disease or NMOSD. The criteria use a "select 6" rule: at least six white matter lesions showing the central vein — or, when fewer than ten lesions are present, more than half of them (2).

Paramagnetic rim lesions (PRLs). These are lesions with a dark rim of iron-laden immune cells around the edge, a sign of smouldering inflammation. They occur in roughly half of people with MS — not sensitive, but very specific. In one multicentre comparison, at least one PRL distinguished MS from mimics and healthy controls with 99.7% specificity and 24% sensitivity (2).

Where they actually change the outcome: with a typical presentation and typical lesions in only one region, a positive select-6 CVS or one or more PRLs, plus dissemination in time or positive CSF, is enough to diagnose MS (2). A small 2026 prospective study at a tertiary centre tested the pair against people who had been misdiagnosed with MS: 66.7% of the confirmed MS group were CVS-positive versus 13.3% of the misdiagnosed group, and 53.3% were PRL-positive versus none (20). Thirty people is a small study, and the authors frame the markers as supportive, not decisive.

CIS, RIS and the other labels you may be given

Not everyone leaves the appointment with a yes or a no. Two in-between labels are common, and in 2024 both moved closer to a diagnosis.

Clinically isolated syndrome (CIS) is a first episode of neurological symptoms lasting at least 24 hours, caused by inflammation and demyelination — optic neuritis, a numb patch, a weak leg — in someone who does not yet meet the criteria for MS. The 2024 criteria note that statements written for CIS also apply to people with a history of relapses who have not yet been diagnosed (2). CSF matters a lot here: OCB-positive CIS carries almost ten times the odds of converting to MS compared with OCB-negative CIS (12).

Radiologically isolated syndrome (RIS) is the incidental discovery of MS-typical lesions on a scan done for something else entirely — a head injury, headaches — in someone with no symptoms or examination findings of demyelination (2). It is not a harmless finding: in the cohorts the committee reviewed, 19% of people with RIS had developed clinical MS by two years, 35% by five years and 51.2% by ten (2).

This is where 2024 broke with every earlier edition. RIS can now be diagnosed as MS when it fulfils dissemination in space plus dissemination in time, or dissemination in space plus positive CSF, or dissemination in space plus a positive select-6 central vein sign (2). The committee calls this a move toward a biological diagnosis — recognising MS by what the tests show rather than waiting for a disabling attack to prove the point (100270-4), 3).

The other tests: eye measurements, evoked potentials and blood work

Optic nerve MRI, OCT and visual evoked potentials. Because the optic nerve is now a fifth location, there are three accepted ways to show it is involved. On MRI, one or more intrinsic optic nerve lesions, with no better explanation, counts. Visual evoked potentials measure how long a signal takes to travel from the eye to the back of the brain; a delayed or asymmetric response supports demyelinating injury. Optical coherence tomography (OCT) photographs the retina's layers: with rigorous quality control, a difference between the two eyes of 6 μm or more in the peripapillary retinal nerve fibre layer, or 4 μm or more in the macular ganglion cell and inner plexiform layer, supports one-sided optic nerve involvement (600275-3)). Optic neuritis is the first manifestation of MS in 25–35% of people with CIS, which is why this change matters (2).

Blood tests. They are mostly there to find something else. There is no blood test that diagnoses MS (7). The 2024 criteria do specifically recommend MOG-IgG testing with a cell-based assay for children under 12 with a first demyelinating event, and for people aged 12 and over whose presentation is atypical for MS or suggestive of MOG antibody-associated disease (2). Most people with neuromyelitis optica spectrum disorder — another autoimmune condition that shares symptoms with MS — have antibodies against aquaporin-4, and blood tests help identify it (18).

Neurofilament light (NfL). You may see this offered. The 2024 CSF panel is explicit: NfL should not be used for diagnosis, though it may be useful for prognosis and monitoring (5).

Why does an MS diagnosis so often take years?

Because the early symptoms are vague, common and easy to attribute to something else — and because the criteria are deliberately cautious. In that US registry of 428 people, the median delay between the first symptom they recognised and the diagnosis was 2 months, but the mean was 22.8 months and the range stretched to 32.9 years. Four in ten said an early symptom had been misattributed. Half reported symptoms they only recognised as MS in hindsight, which had never been assessed — a median of 5.4 years before diagnosis (11).

The specific symptoms that get dismissed are the ones with a hundred other causes. Crushing MS fatigue looks like burnout; brain fog looks like stress and poor sleep; tingling and numbness look like a trapped nerve. Our guide to early MS symptoms goes through which patterns tend to make a neurologist look harder — and if your main complaint is exhaustion rather than anything neurological, what fatigue actually is is the more useful starting point.

Caution also cuts the other way. In a study of 241 new patients referred to two academic MS centres in Los Angeles carrying an established MS diagnosis, 17% at Cedars-Sinai and 19% at UCLA turned out to be misdiagnosed — nearly one in five. The most common correct diagnoses were migraine (16%), radiologically isolated syndrome (9%), spondylopathy (7%) and neuropathy (7%), and the misdiagnosed group had received around 110 patient-years of unnecessary MS treatment between them (10). That is the cost the criteria are trying to avoid, and it is why your neurologist is in no hurry to say yes.

For scale: an estimated 727,344 US adults were living with MS in 2010, with a prevalence of 309.2 per 100,000 and a female-to-male ratio of 2.8 (14).

What else can look like MS

"No better explanation" is doing heavy lifting in the criteria, and this is the list it is pointed at:

• Migraine with white matter lesions — the single most common alternative diagnosis in the Los Angeles misdiagnosis study (10).

• Small vessel disease from blood pressure, diabetes, smoking or age — the reason for the extra safeguards at 50 and over (2).

• NMOSD and MOG antibody-associated disease — both cause optic neuritis and spinal cord inflammation, both are identified with antibody testing, and both are managed differently (2, 18).

• Transverse myelitis from another cause — spinal cord inflammation that develops over minutes to days or over one to four weeks, sometimes the first sign of MS and sometimes not (16).

• Spinal problems and nerve compression — spondylopathy and neuropathy together accounted for 14% of corrected diagnoses (10).

• Conditions that overlap symptomatically rather than on MRI — widespread pain and exhaustion send plenty of people down an MS pathway; we cover how clinicians work through testing for fibromyalgia and how EDS and hypermobility are diagnosed separately.

Treatment is a different subject entirely, and not one for this page: disease-modifying therapy decisions belong with your neurologist once the diagnosis is settled.

Living with an MS workup: the questions people ask next

My MRI showed lesions but my neurologist won't say MS. Why not?

Because lesions alone are not the criteria. The diagnosis needs typical lesions in at least two of five locations plus either dissemination in time, positive CSF, or — in specific situations — supportive markers, and it needs no better explanation for any of it (2). If your lesions are in one location, or sit in a pattern more typical of migraine or small vessel disease, more evidence is genuinely needed. The 2024 criteria also say the diagnosis should be reassessed periodically, which is why "let's scan again in six months" is a plan and not a brush-off (2).

Can I refuse the spinal tap?

You can decline any test, and sometimes the diagnosis is reached without one — four of five locations with a typical presentation needs nothing further (2). But when the MRI is borderline, positive CSF is often the thing that converts a maybe into a yes, and it also screens for infections and other inflammatory conditions (7). If the headache risk is the sticking point, ask whether the clinic uses atraumatic needles (13).

My scan was ordered by a GP. Does that count?

Often not for diagnostic purposes. MS protocols call for brain and spinal cord coverage, fat-saturated sequences for the optic nerves where relevant, and susceptibility-sensitive sequences for CVS and PRLs (400304-7)). A general brain MRI may lack all of these. Ask whether the images you already have are adequate for MS criteria or whether a protocol scan is needed.

I'm 56 and they keep adding tests. Is that normal?

Yes, and it is written into the criteria. For people being considered for MS at age 50 or over, or with hypertension, smoking, diabetes, high cholesterol, known vascular disease or a headache disorder, the committee strongly recommends additional confirmation — a spinal cord lesion, positive CSF, or a positive central vein sign — before the diagnosis is made (2).

I was diagnosed years ago under older criteria. Does the 2024 version change anything?

Not retroactively, but it is a reasonable thing to raise, particularly if your original workup was thin or your course has not behaved as expected. Misdiagnosis is common enough — close to one in five in a referral population at two academic centres — that MS specialists revisit established diagnoses routinely (10), and the criteria themselves call for periodic reassessment (2).

My clinic doesn't offer central vein sign or kFLC. Am I getting worse care?

Not necessarily — neither is required for a diagnosis, and both exist to add specificity in ambiguous cases (2). Access is genuinely uneven: in a 2026 survey of 24 MS centres in Portugal, CVS and PRLs were not routinely analysed in 37.5% and the kFLC index was routine in 41.7% (22). If your case is borderline, asking for a referral to a centre that does offer them is a fair request.

How to bring this up with your doctor

Bring a dated timeline, not a feeling. The question a neurologist is trying to answer is whether anything happened before, and when. Write down each episode: what you noticed, which body part, the date it started, how long it lasted, and whether it fully resolved. Episodes lasting at least 24 hours matter; so does anything you dismissed at the time. Half of the people in that registry had symptoms nobody ever evaluated (11).

What to say. "Over the last [X] years I've had these separate episodes, on these dates, each lasting [X] — here's the list. Is a demyelinating process worth ruling out?"

What to ask.

• "Does my MRI cover the brain and the spinal cord, and was it done to an MS protocol?"

• "Which of the five locations show typical lesions?"

• "Is dissemination in time established, or do we need a repeat scan?"

• "Would CSF testing — oligoclonal bands or the kappa free light chain index — change your thinking?"

• "Have NMOSD and MOGAD been excluded?"

• "If this isn't MS, what is the working diagnosis, and what's the plan for it?"

Bring your records. Prior MRIs on disc, not just reports; previous eye exams if you have ever had optic neuritis; a list of your vascular risk factors, since they change what the criteria require of you (2).

Ask for an MS specialist if you are stuck. The 2024 criteria require clinical and radiological expertise in MS to apply well — the committee lists this among the challenges of its own revision (2).

The red flags, which override all of the above. Call 911 for sudden numbness or weakness of the face, arm or leg (especially on one side), sudden confusion or trouble speaking, sudden trouble seeing, sudden trouble walking or loss of balance, or a sudden severe headache with no known cause — these are stroke warning signs. Get emergency care the same day for new weakness in both legs, a band of numbness or tightness around your torso, or new trouble controlling your bladder or bowels. Get urgent care for new eye pain with loss of vision in one eye (15, 16, 17).

Start understanding your body

Learn what affects your energy, stress, sleep, and daily state. Get the app.

Explore your data

How Welltory helps — and what it cannot do

The limits first, because they are absolute here. Welltory is a general wellness product, not a medical device. It does not diagnose, detect, predict, monitor, prevent, treat or mitigate any condition. It cannot see multiple sclerosis — not lesions, not demyelination, not an attack, not a remission. There is no metric in the app, and no combination of metrics, that says anything about whether you have MS. Nothing in your Welltory data belongs in the McDonald criteria, and no neurologist will use it as evidence. Welltory's help centre also notes that measurements may not be accurate if you are under 18 or pregnant, because those body systems work differently and the algorithm is less reliable in those cases (21).

We also cannot show you a Welltory chart of "what MS looks like in the data," and we are not going to invent one. We have no cohort of people with a confirmed MS diagnosis, no before-and-after around a diagnosis date, and no reason to believe such a chart would mean anything if we drew it. A health product that cannot detect a disease should not publish a picture implying otherwise.

What it can honestly do is keep a dated record while you wait. The gap this fills is the one described above: from a first symptom to a diagnosis, the mean delay in a US registry was 22.8 months, and half of participants had earlier symptoms nobody evaluated, a median 5.4 years before diagnosis (11). A neurologist asks "when did this start, and how long did it last?" and most people cannot answer from memory. A dated note can.

You tag the event; the app records the rest the same way every day. Add a tag from the plus sign at the top of the feed — "left eye blurry, day 3", "numb patch on forearm", "legs heavy after a hot shower". When the Today screen marks a stress or rest stretch, it asks "What happened?"; tap a suggested tag, type or talk, and correct the tags before saving.

The details screens are where the record lives. Resting heart rate, heart rate variability, stress minutes, Battery and sleep are measured the same way each day, so your readings sit against your own recent history rather than any reference line. Battery and stress minutes need iOS with an Apple Watch or Oura; detailed sleep analysis needs an Apple Watch. None of those numbers is an MS signal. What they are is a consistent backdrop that makes "this week was different from the three before it" checkable rather than remembered.

Over a few months, My Patterns shows what your tags have in common. My Patterns collects tags from stress and rest stretches on iOS with an Apple Watch or Oura, and patterns start appearing at around seven tagged events in a month. Any pattern it finds is an association, not a cause — and certainly not a diagnosis.

A record you made, for an appointment. The Journal (Premium) keeps your measurements, tags and notes, and from the web app you can export a CSV (Dashboard → choose a chart → Export). A dated line like "18 March, right eye painful on movement, blurred for 11 days, resolved" is exactly the kind of thing the McDonald criteria care about and memory is terrible at.

The same red flags apply here, in the same type size. Call 911 for sudden numbness or weakness of the face, arm or leg (especially on one side), sudden confusion or trouble speaking, sudden trouble seeing, sudden trouble walking or loss of balance, or a sudden severe headache with no known cause — these are stroke warning signs. Get emergency care the same day for new weakness in both legs, a band of numbness or tightness around your torso, or new trouble controlling your bladder or bowels. Get urgent care for new eye pain with loss of vision in one eye (15, 16, 17).

How we made it

Made with AI tools, then edited and fact-checked by the Welltory team. See our Editorial & AI policy.

Jane Smorodnikova, co-founder of Welltory, built the methodology behind how we read physiological data.

Written by Tatsiana Yashyna.

Discounts for blog readers: up to 36% off

See what affects your energy, stress, sleep, and daily state with Welltory

Explore your data