Art Therapy for Trauma and PTSD: What the Controlled Evidence Supports

Art therapy for trauma and post traumatic stress disorder is psychotherapy in which image making is part of the work, and on this topic the claims made for it run a long way ahead of the trials behind them. The gap is not a scandal and it is not evidence of failure. It is the ordinary consequence of a small profession working with a population that is hard to study, in services that fund delivery rather than research. What makes this page different from most is that it also has to say something advocacy pages leave out: this is one of the areas where the work, delivered badly, can leave somebody worse than it found them.

I have no first hand account of this and it would be wrong to invent one. What I have instead is the reason I keep saying no. Two veterans’ groups and a refugee support project have asked me, at various points, to run something they described as trauma singing. It is a real and understandable request: singing together does something to a room, people who cannot talk about an experience will sometimes sing near it, and I am cheap and available in a way that a registered therapist is not. I decline every time, and the reasons are not modesty. I have no training in working with traumatic material, so I would not recognise dissociation if it happened in front of me and I would not know what to do about it. I have no supervision and nobody clinical to escalate to. My sessions end at four o’clock and everybody goes home, which is a fine arrangement for a singing group and an appalling one for somebody who has just been opened up. And I have watched enough enthusiasm in this sector to think this is the one area where enthusiasm is genuinely dangerous. I am not the person to run it. What follows comes from the profession’s published description of its own practice, from the reviewer who checks this page, and from the trials that exist, which are not about this.

The tier labels used throughout this site (Tier 1 supported, Tier 2 promising but limited, Tier 3 practice and anecdote) matter on this page mostly where they are empty, and what an empty one does and does not mean is set out on reading arts in health research.

What is claimed, and where the claim comes from

The argument you will meet everywhere is that traumatic memory is held non verbally, so image based work can reach material that talking cannot. It is a rationale. It is not a finding, and the difference is the whole subject of this section.

Stated carefully, the rationale runs like this. People who have been through overwhelming experiences frequently cannot produce a coherent verbal narrative of them, either because the memory does not present itself as a story or because saying it out loud is intolerable. Making an image does not require a narrative, can be approached obliquely, and gives a person something outside themselves to look at. Therefore, the argument concludes, image based work may reach people whom talking therapies do not.

Every step of that is plausible and none of it is evidence. It explains why practitioners thought this would work. It does not establish that it does, and it certainly does not establish that it does better than a verbal therapy, which is a further claim again.

How a mechanism becomes a finding

Watch the sentence change shape. This is the single most useful thing on the page, because once you can see it happening you can see it happening in every arts in health topic.

It starts as a hypothesis in a practitioner’s account: image making may reach material that is not available to language. In the next document it becomes a description of practice: art therapy works with material that is not available to language. In the one after that, the modal verb has gone and the claim is about people: art therapy reaches trauma that talking cannot. By the fourth repetition it is in a funding bid as a benefit, and by the fifth it is in a newspaper as something research has shown. At no point did anybody lie, and at no point did a study enter the chain.

Two tells are reliable. The first is a claim with no design attached: no comparator, no outcome measure, no sample, no year. The second is a citation that leads to another citation. If following a reference takes you to a review that cites a book chapter that cites a case series, the figure has been forwarded rather than checked, and that happens in trauma writing more than almost anywhere, because the underlying case material is genuinely compelling.

What the work involves when it is done properly

Properly run trauma work with a registered therapist is slower, duller and far more structured than the version in the public imagination, and most of it is not about the traumatic material at all.

The shape most services describe is phased. The first phase is stabilisation: establishing that the room is predictable, that the person can stop, that the therapist will not push, and building what practitioners call resources, which means practical ways of settling when something surfaces. That can occupy a great deal of the work, and in some cases it is the entire work, quite legitimately. Only after that does anything approach the material itself, at a pace the person controls, and often obliquely rather than head on.

“Draw the worst thing that happened” is the wrong model, and it is the model most people picture. Producing a graphic reconstruction of an event, in a room with no containment around it and nobody to help afterwards, is exposure stripped of every structural element that makes exposure a treatment rather than a repetition. It can leave a person flooded, dissociated, or more symptomatic for days. The prompt sounds therapeutic and is closer to a re enactment.

What a trained practitioner does when someone destabilises in the room is where the training shows. Broadly: stop the approach to the material, bring attention back to the present and to the physical room, use whatever grounding the two of them established earlier, slow everything down, and spend the remaining time settling rather than continuing. Then attend to what happens next, which means knowing whether the person is safe to leave, what support exists tonight, who else needs to know, and what happens at the next appointment. That last part is invisible to a participant and it is most of what distinguishes clinical work from a workshop. Around all of it sits the ordinary clinical scaffolding: assessment, written goals, notes, clinical supervision and a planned ending, described on what happens in an art therapy session, and the consent questions that come with working in wards, prisons and hostels are on consent and safeguarding in arts in health.

The evidence position, stated plainly

No effect size for art therapy in trauma or PTSD appears anywhere on this site, and the blank is chosen rather than accidental.

No result in this area has been checked against its source document by the reviewer who signs off every figure here. That is the whole reason. Tier 3, of the not verified here kind, and not the null kind. Nothing on this page is a finding that art therapy fails to help people with trauma. There is no such finding here to report, which is a different statement, and the two get written identically almost everywhere.

What would have to exist for a number to appear:

• a controlled trial in people with trauma related symptoms, with the comparator named
• a blinded rater, because an outcome scored by somebody who knows who received what is the weakest link in this entire literature
• a named PTSD measure, specified in advance rather than chosen from among several afterwards
• an adequate sample, powered for the effect being looked for rather than for whoever turned up
• follow up beyond the last session, since an effect present in the final week and gone by the third month is a different claim
• a certainty rating, or an explicit statement that the review predates routine certainty rating
• a year, and a source anyone can open

Or a systematic review pooling trials that meet those conditions. Nothing less than that gets printed here, on this topic or any other.

It is also worth saying what this page is not doing. There are established psychological treatments for PTSD with much larger controlled literatures behind them than anything in the arts therapies, and someone seeking treatment should be asking their clinician about those first. This site does not grade them, because they sit outside what has been verified here, and an arts page is the wrong place to learn about them.

The borrowing trap

You will meet numbers in trauma contexts that were produced somewhere else entirely. Two of them are locked figures on this site, which makes them worth naming so you can recognise them when they are handed to you as though they were about trauma.

MATISSE. A three arm, rater blinded, pragmatic randomised controlled trial in secondary care across four UK centres. 417 people randomised: 140 to group art therapy, 140 to activity groups as an active comparator and 137 to standard care alone, with 355 (85%) followed up at 2 years¹. Against standard care at 24 months, global functioning came out at an adjusted mean difference of -0.9 (95% CI -3.8 to 2.1) and symptoms at 0.7 (95% CI -3.1 to 4.6). No difference between the three arms on either primary outcome, and no secondary outcome favoured art therapy; the single secondary difference ran the other way, with the activity group arm having fewer positive symptoms at 24 months. Attendance at the art therapy groups was not associated with improvement, and only 86 of the 140 randomised to art therapy (61%) attended even one group². Tier 3, of the null controlled evidence kind, not the nobody has looked kind.

That result is about referral to twelve months of weekly group art therapy, as an adjunct, for adults with established schizophrenia, on those two outcomes. It says nothing about individual work, nothing about other diagnoses, nothing about first episode psychosis, and nothing at all about trauma. Used as a general verdict on art therapy it is a misreading, and used as a verdict on trauma work it is not even the right question. The full account is on art therapy for psychosis.

The 2021 Cochrane review of music interventions in people with cancer. 81 trials, 5,576 participants, anxiety 7.73 STAI-S units lower, pain SMD -0.67, both at very low certainty³. Tier 2 for both. It is a music review, in an oncology population, and it turns up in trauma material because it is one of the few large pooled arts figures in existence and because eighty one trials sounds decisive. It is not transferable in either direction, and its own reviewers rated their confidence in it as very low. Both of these are averages across large groups and neither predicts what will happen to one person.

The general habit worth acquiring: when a figure appears, ask which population, which intervention, which outcome, and which year, before asking how big it is. A number that answers a different question is not weak evidence for this one. It is not evidence for this one at all.

The risk, which advocacy pages leave out

In most of arts in health the realistic downside of a badly run session is a wasted hour. Here it is not, and that asymmetry is the reason this section exists.

The specific failure modes are known and they are ordinary rather than exotic. A person is invited to approach material before there is any stability to hold it, and leaves flooded. A prompt intended to be expressive functions as a demand and produces dissociation, which an untrained facilitator may read as calmness. A group is convened in which one participant’s disclosure destabilises three others in the same hour, with one person running the room. Something surfaces in the last ten minutes and there is nowhere for it to go, because the session ends and the building closes. Somebody deteriorates in the week afterwards and there is no route to anyone clinical, no notes, and no one who knows they were there.

None of that requires bad intentions. It mostly requires enthusiasm, a room and no framework. It is also why a phrase like trauma informed, used as a badge on a workshop listing, tells you nothing on its own.

Which is why the register check matters more on this topic than on any other page here. Four arts therapy titles are protected in UK law: art therapist, art psychotherapist, dramatherapist and music therapist. The HCPC register listed 6,103 arts therapists across all four in July 2026⁴, and using one of those titles without being registered is a criminal offence. Registration means a postgraduate qualification with supervised clinical placements, continuing professional development, and accountability to a regulator that can remove someone from the register, which is the route described on training as an art therapist. The professional body publishes practice standards and maintains a members directory⁵, and membership is not the same as statutory registration; check both. Someone unregistered is not necessarily unskilled, and the point is not their competence but the fact that nothing sits behind them if it goes wrong.

Getting to it in the UK, and what to ask

Routes are narrow. Arts therapists working with trauma are found in secondary mental health services, in specialist trauma and complex needs teams, in services for refugees and survivors of torture, in some veterans’ services, in forensic settings, and in the voluntary sector, plus private practice. Provision is patchy and depends on whether a post exists, and arts in mental health inpatient units covers what the work looks like on a ward, where the constraints are different again.

Questions worth asking before you start, and none of them is rude:

1. What training does this practitioner have in trauma specifically, as distinct from art therapy generally, and are they on the HCPC register.
2. What does the plan look like for the first few sessions, and does stabilisation come before any approach to the material.
3. What happens if I become very distressed in the room. What happens after I leave.
4. Who can you escalate to, and how fast.
5. Is this individual or group, and if group, how are disclosures handled and who else is in the room.
6. What is written down, who reads it, and what happens to anything I make.
7. What if it goes badly and I want to stop. Is stopping a decision we make, or is it recorded as me dropping out.

Question seven is the one people skip and the one that tells you most. A service treating departure as a clinical decision to be discussed is a service that expects to be wrong sometimes. If work does go badly, say so to the practitioner directly, raise it with whoever referred you, contact the regulator if the person is registered, and tell your GP, because the aftermath is a clinical matter rather than a customer service one.

Three related routes sit in the same area with different structures. Dramatherapy in mental health services covers a fourth protected title that works through role and distance, which some people find easier than an image. Therapeutic writing and photography covers practices outside the register entirely, which carry the same risks without the same accountability. Music therapy for depression and anxiety covers an adjacent population, and art therapy is the pillar for the profession.

Now the caution in both directions, and it needs stating hard. An empty evidence row on this page is not a verdict. It records that nothing has been verified here to a standard worth printing, and it says nothing whatsoever about whether this work helps: a great many people describe art therapy as the thing that finally reached something years of talking had not, and that testimony is real, even though testimony is not a trial and cannot be turned into one by volume. What nobody can honestly do is promise you it will help, and what nobody can honestly do is tell you it will not. What can be said with confidence is narrower and more practical: the person delivering it should be trained for this specifically, accountable to somebody, and able to tell you what happens when it goes wrong. And whatever this work turns out to be worth for a particular person, it belongs alongside the treatment they are receiving rather than instead of any part of it.