Rhythmic Auditory Stimulation: The Strongest Evidence in Arts and Health

Rhythmic auditory stimulation is a gait training technique in which a steady external beat, usually a metronome or a piece of music with a strong pulse, is set to match the cadence of somebody’s steps and then moved towards a target, so that the walking follows the beat rather than the beat following the walking. After stroke it has the best controlled evidence of anything described on this site. It also carries the narrowest claim on this site, and those two facts together are why it has a page of its own rather than a paragraph on music therapy.

I first saw it done through a gap in a curtain. I run a weekly singing and songwriting session on a stroke rehabilitation unit, as a community practitioner rather than a registered therapist, and the physiotherapists there sometimes work with a metronome in the corridor outside the bay I use. What I watched was duller than I expected. A click set at what looked like an uncomfortably slow rate, one physiotherapist counting a man in, a second at his elbow, a short walk and back, then the click moved up by an amount I could not hear and the same walk again. Nobody was enjoying it. It looked like practising scales.

Some weeks later a nurse who had seen both that corridor work and my session asked me whether I could do the rhythm thing on a Thursday, for a man who was struggling to walk. I said no, and it took me longer than it should have to explain properly why. What the physiotherapists were doing was not using rhythm. They had measured his cadence, picked a target, matched the beat to the first and were walking it towards the second, they were watching his hip and his weak side the entire time, and there was a second pair of hands within reach in case he went. I have a keyboard, a folder of songs and no business setting a gait target for anybody. The two activities share a pulse and nothing else that matters. The request revealed how quickly a specific clinical technique gets read as a general musical one, which is a mistake nobody would make about a drug, and it happens here because the visible ingredient is a beat and everybody recognises a beat.

This page states one Tier 1 claim and then spends most of its length refusing to let it spread. The labels used across the site are Tier 1 supported, Tier 2 promising but limited, Tier 3 practice and anecdote, and what a claim has to show to earn each of them is set out on reading arts in health research.

What rhythmic auditory stimulation actually is

It is an external beat used as a timing target for stepping, on the basis that the motor system will couple to a periodic sound whether or not the person is trying to make it do so.

That coupling is what people mean by entrainment, and you can observe it in yourself without equipment: walking beside somebody, your steps tend to fall into theirs. In rehabilitation the same tendency is put to work in a specific order. Somebody measures how fast the person is currently stepping. The beat is set to that, so the walking and the sound agree from the start. Then the beat is moved, in small increments, and the walking is asked to follow. Cadence is the direct target; stride length, symmetry between the two sides, and the time spent on the weak leg are usually being watched alongside it.

Three features of that sequence do the work, and none of them is the music. The beat is set against a measurement rather than chosen for pleasantness. It changes on purpose. And somebody is present who can see what the body is doing to keep up, because the wrong way to walk faster is to take the same steps in less time by rushing and leaning.

So this is not listening to music while you walk. It is not a playlist for the treadmill, not background music in a gym, and not a rhythmic exercise class. Those may all be worth doing and none of them is what the trials on this page tested.

The finding, in full

The Cochrane review of music interventions for acquired brain injury reports that rhythmic auditory stimulation improved gait velocity by 11.34 m/min compared with the control conditions, with a 95% confidence interval of 8.40 to 14.28, pooled from 9 trials and 268 participants, at moderate quality¹. That is Tier 1.

Review	Cochrane CD006787.pub3, 2017
Outcome	Gait velocity
Result	11.34 m/min faster
95% confidence interval	8.40 to 14.28
Trials	9
Participants	268
Quality of evidence	Moderate
Tier on this site	Tier 1

Read the interval before the estimate, because it is the interval that makes this unusual. It does not cross zero, and the end of it least favourable to the technique is still an improvement of a size worth having. That is a different situation from most of the findings quoted on this site, where the interval either straddles zero or is so wide that the estimate in the middle of it is doing more rhetorical work than statistical work.

The quality rating matters as much. Moderate quality means the reviewers thought the pooled answer was reasonably trustworthy given how the individual trials were run, and in this field a rating that high is rarer than a large sample. It is worth saying plainly that 9 trials and 268 participants is a small evidence base by the standards of, say, a cardiology review. It is a large one by the standards of arts in health, and it is the only place on this site where all five things a Tier 1 label requires are present without argument: an effect size, a confidence interval, a quality rating, a sample size and a year. Everywhere else I have to explain which of the five is missing and why.

What gait velocity is, and what a group average will not tell you

Gait velocity is simply how fast somebody walks over a measured distance, and rehabilitation teams treat it as a summary measure because so much else is folded into it.

Balance, strength on the affected side, confidence, coordination and fear of falling all show up in walking speed, which is why it gets tracked when a more specific measurement would be harder to take and harder to interpret. It also has an unglamorous practical importance that the research summaries leave out. Walking speed is the difference between crossing a road inside the time the lights allow and not; between reaching a toilet and not; between walking with your family and being waited for.

The figure is nonetheless a difference between group averages, and that limitation is not a formality. Inside those 9 trials some participants improved a great deal, some improved not at all, and some will have got slower over the study period for reasons unconnected to the intervention. A group mean difference is not a prediction about a person. If a stroke team offers this to somebody you know, the honest expectation is that it has a good chance of helping their walking speed and no guarantee, which is a weaker sentence than the number invites and a truer one. It is also a difference measured at the point the trials measured it, and how long an improvement lasts after a course ends is a separate question this number does not answer.

The unit, and why it stays as it is

The review reports metres per minute. This site quotes metres per minute, in every article, without exception.

The reason is not pedantry. Conversion is one of the two or three places where an honest writer accidentally improves a result. The same difference stated in centimetres per second is a different looking number carrying exactly the same information, and if the only thing a change of unit can add is an impression, then the impression is the reason it was made.

There is a second and more practical reason, which is auditing. If one figure on a page has been restated for effect, a reader has no way of knowing which of the others have been. The same discipline governs the rest of the site: standardised mean differences do not become percentages here, and nested study counts are not laid out side by side as though they were additions. The reasoning is set out at greater length in how to read a figure in this field.

How narrow this claim is

This is a result about walking speed. It is not a result about mood, communication, confidence, independence or quality of life after stroke, and the tier attached to it grades that single outcome and nothing adjacent to it.

I labour this because the pressure runs the other way. This is the best number the field has, and a sector that spends much of its energy arguing for funding has an obvious use for a strong result: attach it, in a general way, to the proposition that music helps people recover from stroke. Every step of that move is small and the destination is a claim the evidence cannot support. If somebody wants to know whether music work after stroke improves mood, the answer has to come from evidence about mood, and on this site there is no locked effect size for it at all. The wider picture, outcome by outcome, is on music therapy after stroke.

The same discipline has to apply when a finding wants to spread in the other direction, and the clearest example is the 2025 Cochrane review of music based therapeutic interventions in dementia, which reports agitation and aggression at SMD -0.05 (95% CI -0.27 to 0.17) at moderate certainty². That is Tier 1, a null result: moderate certainty evidence of no meaningful effect on agitation, which is a real finding rather than a gap. It is also a finding about one outcome in one population, and it does not license a general conclusion that music does nothing in dementia. Read carelessly in either direction, a narrow result becomes a broad one, and the broad version is wrong both times. The rest of that review, including the row where the effect is real, is on what the dementia evidence shows.

Who delivers it, and who it is likely to suit

Most often a physiotherapist. Sometimes a music therapist with additional training in neurological rehabilitation techniques. On units where both are present, occasionally the two together, which is the arrangement the technique was designed around.

The overlap between the two professions here is genuine and slightly awkward. Gait is unambiguously physiotherapy territory, and a physiotherapist has the assessment skills, the handling skills and the professional footing to be the person walking beside somebody who might fall. A music therapist brings a different competence: playing music that holds a pulse reliably while a person is struggling, adjusting it live rather than pressing a button, and doing that without the music becoming the event. Where a unit has one rather than both, it is nearly always the physiotherapist, and a metronome application on a telephone costs nothing.

That is not an accident of preference, it is a workforce fact. In July 2026 the Health and Care Professions Council register listed 6,103 arts therapists across all four protected titles, for the whole United Kingdom and every setting in it³. Spread that across hospitals, schools, hospices, prisons, community teams and private practice and the number of stroke units with a music therapist attached is small. The British Association for Music Therapy is the place to look for who is practising and where⁴, and the practical routes are set out on how to find a music therapist.

The people this suits best are those who are already taking steps, with or without support, who can hear and attend to a beat, and who are at the stage of rehabilitation where walking is an active goal. That describes a large proportion of people on a stroke rehabilitation ward and a smaller proportion of people in the acute days immediately after a stroke. What a music therapy session looks like more generally, when it is not aimed at gait, is described on what happens in a music therapy session.

Who it is unlikely to suit

People who are not yet walking, people who cannot reliably perceive or hold on to an external beat, and people whose principal difficulty after stroke is not their walking.

Hearing loss is the most straightforward exclusion, and it is common in the age groups most affected by stroke: if the beat is not clearly audible it is not a target. Difficulties with attention, or with awareness of one side of the body, can also make an external cue hard to use, because following a beat is a sustained attentional task as well as a motor one. And some people simply find a metronome unpleasant to walk to, which is a good enough reason on its own.

Then the largest group: people for whom walking is not the problem. If somebody’s stroke has left communication as their dominant difficulty, the strong number on this page is not about them, and saying so is more useful than offering it as encouragement. Anything delivered on a ward also runs into questions about who agreed to it, and what capacity to consent looks like when somebody’s communication has been affected, which is set out on consent and safeguarding in arts in health.

How this sits beside the dance and movement work

The equivalent figure does not exist for dance based movement work after stroke, and that absence is a statement about the research literature rather than a verdict on the activity.

I am unusually well placed to notice the gap, because a seated dance class uses the hall on the unit immediately before my session does. It uses rhythm. It involves movement. Participants like it, and I have watched people move in it who do very little in the rest of the week. There is no effect size on this site for it, and that is a decision rather than an omission. For one to appear here, a systematic review or an adequately powered randomised controlled trial would have to report a named outcome with an effect size, a confidence interval, a certainty rating, a sample size and a year, readable end to end rather than through a report quoting it. That does not exist in a form this site is willing to lock, so dance work after stroke sits at Tier 3, of the kind where nobody has yet looked adequately rather than the kind where the controlled evidence is null. Those two situations get labelled identically almost everywhere, and the difference is the whole reason for saying which one is meant. See dance after stroke, the wider map on dance for health, and dance, balance and falls, where the same absence applies.

Part of the reason is that rhythmic auditory stimulation is unusually easy to trial. One clear outcome, measured with a stopwatch rather than a questionnaire, and a control condition that is straightforward to arrange. A seated dance class has none of those properties: it is a bundle of ingredients aimed at several outcomes at once, most of them measured by asking people how they feel. That is a reason the evidence is thinner, and not a reason to conclude anything about the class.

What to ask a stroke team

The single most useful question is whether rhythmic cueing is already part of the gait work, and if not, whether it has been considered.

Four more that get better answers than a general request for music:

• Which outcome is being tracked at the moment, and is walking speed one of them? A team that is measuring something can tell you whether it is moving.
• Who would deliver it here, the physiotherapist or a music therapist, and does this unit have the second?
• Is this something to continue after discharge, and if so, who sets the target and reviews it?
• What is available once rehabilitation ends? This is the question with the worst answer nearly everywhere, and it is better to ask it early than to discover it on the day.

And one thing not to do: do not ask for music therapy if what you want is gait cueing. They are different requests, addressed to different professions, with different evidence behind them, and a muddled request tends to be met with whichever of the two the ward happens to have. Music programmes inside a hospital are a third thing again, and not a clinical service at all: see arts in hospitals.

What I find most useful about this finding is not its size but its shape. One well specified technique, one measurable outcome, one population, and a number that survives being checked. Very little else in this field looks like that, and the reason is usually not that the activities are useless but that nobody has asked a question narrow enough to answer. And none of it is a reason to alter or decline anything a stroke team has arranged: a beat that helps somebody walk faster is an addition to their rehabilitation, never a substitute for any part of it.