Is IFS Evidence-Based? The Research, the Criticism, and the Honest Answer

If you search for Internal Family Systems right now, you will not mainly find explanations of what it is. You will find an argument about whether it is real. Magazine investigations calling it dubious. Psychiatrists quoted describing it as, at best, a simplistic allegory. Practitioners responding that the critics have never sat in a session. Somewhere underneath all of it is a question most people actually want answered, and almost nobody answers straight: what does the research say?

This is our attempt at the straight version. We build IFS-informed tools, so we have an obvious interest in the answer, and that is exactly why the numbers below are cited rather than characterised. Where the evidence is thin, we say it is thin. Where the critics are right, we say they are right.

What the evidence base actually contains

The most useful single document here is recent. In 2025, M. E. Buys published a scoping review in Clinical Psychologist titled "Exploring the evidence for Internal Family Systems therapy: a scoping review of current research, gaps, and future directions." A scoping review is the right instrument for a young literature. Rather than pooling effect sizes from studies that may not be poolable, it maps what exists.

Twenty-seven studies met the inclusion criteria. The breakdown is the part worth remembering:

Study design	Count	What it can establish
Case studies	17	That something happened, and how it looked up close
Quasi-experimental	5	Change over time, without random assignment
Qualitative exploratory	3	What the experience is like, in participants' own words
Randomised controlled trials	2	That the therapy, rather than time or attention, caused the change

Seventeen of twenty-seven are case studies. Case studies are not worthless, and the history of psychotherapy would be empty without them, but they cannot tell you whether a person would have improved anyway. They are where a field starts, not where it arrives.

The review's positive findings are real and should not be buried under the methodology. It describes IFS as a promising treatment, particularly for chronic pain, depression and post-traumatic stress disorder, and for developing self-compassion and self-forgiveness. It then names explicit gaps in feasibility, acceptability and effectiveness, which is a polite way of saying that the basic questions have not been answered yet: whether the therapy can be delivered as intended, whether people will stay in it, and whether it works better than the alternatives.

The two randomised trials, and the one that is not

Because the trial count is two, it is worth knowing which two, and worth knowing what gets miscounted as one.

The best known is Shadick and colleagues (2013), published in The Journal of Rheumatology: a proof-of-concept randomised trial of an IFS-based intervention in rheumatoid arthritis. Seventy-nine patients were randomised, thirty-nine to nine months of IFS and forty to a mailed-education control, with joint counts assessed by rheumatologists blinded to group assignment. It reported reduced pain and depressive symptoms, and improved physical function and self-compassion. The blinded-assessor design is genuinely good practice and the finding is a real one.

Note what it is, though. It is a proof-of-concept study in an autoimmune disease population, not a mental health efficacy trial. That becomes important in the next section.

The study most often cited as IFS's trauma evidence is Hodgdon and colleagues (2021), in the Journal of Aggression, Maltreatment & Trauma, on IFS for PTSD among survivors of multiple childhood traumas. It found significant positive effects. It is also an uncontrolled pilot with seventeen participants. There was no comparison group. It is meaningful preliminary work, and it is routinely described online as though it were a randomised trial, which it is not. If you have seen IFS called an evidence-based PTSD treatment, this seventeen-person pilot is usually what is doing the work.

The SAMHSA listing, honestly explained

Almost every defence of IFS reaches for the same sentence: IFS is listed by SAMHSA as an evidence-based practice. It is true, and critics are right that it carries less weight than the phrasing implies.

Here is the full picture. In November 2015, IFS was added to SAMHSA's National Registry of Evidence-based Programs and Practices, following an application submitted in April 2014. The independent review rated it effective for improving general functioning and well-being, and promising for reducing depression, anxiety and physical health symptoms and for improving personal resilience.

Three qualifications belong with that sentence every single time it is used.

First, the underlying study was the rheumatoid arthritis trial. The federal listing that gets cited as proof IFS works for trauma rested substantially on a proof-of-concept study in people with an autoimmune joint disease.

Second, the registry took voluntary submissions and said so. Programs were not swept up by systematic review; their developers applied. The registry's own homepage stated that inclusion did not constitute an endorsement. A listing meant a submission had cleared a methodological bar, not that a federal agency was recommending the treatment.

Third, and most decisively, the registry no longer exists. SAMHSA phased NREPP out in 2018 and replaced it with the Evidence-Based Practices Resource Center, which explicitly does not carry over everything NREPP contained. Citing the registry today means citing a database that has been dark for years.

How this compares to CBT and EMDR

Scale is the thing people misjudge. Cognitive behavioural therapy has hundreds of randomised trials and meta-analyses of meta-analyses across dozens of diagnoses. EMDR, itself once treated as fringe, accumulated enough trials to earn treatment-guideline recommendations for PTSD from multiple national bodies. IFS has two randomised trials.

That is not a rhetorical flourish, it is the actual ratio, and it should change how you talk about the model. IFS is roughly where EMDR was decades ago rather than where EMDR is now. We wrote a fuller comparison of IFS and CBT that goes through what each approach is actually trying to do, and the evidence gap is the honest headline of that comparison too.

It is also worth saying plainly that a thin evidence base is not evidence of absence. Psychotherapy research is expensive, slow, and structurally biased toward manualised short-term treatments that are easy to standardise and cheap to trial. Approaches that are relational, longer, and harder to manualise get studied less regardless of whether they work. That explains some of the gap. It does not close it, and "it is hard to study" is not a finding.

The criticism, stated properly

Criticism of IFS is frequently strawmanned by its defenders, which is a mistake, because the strongest version is not about whether talking to your parts feels useful. It is about the gap between reach and evidence.

The popularity-to-evidence mismatch

The most rigorous critique came in June 2024 from Lisa M. Brownstone, Madeline J. Hunsicker and Amanda K. Greene, writing for the Society for the Advancement of Psychotherapy under the title "Internal Family Systems: Exploring Its Problematic Popularity." Rather than arguing from impressions, they counted. As of April 2024:

• 45,764 therapists listed IFS as a service on Psychology Today
• 74,154 Instagram posts tagged #internalfamilysystems, plus 30,875 tagged #ifstherapy
• Roughly 3 million TikTok posts mentioning Internal Family Systems therapy

Set 45,764 practitioners against 27 published studies and the shape of the problem is obvious without any further argument. Their conclusion is that the expansion of IFS has moved beyond its evidence base, and on the arithmetic they are simply correct.

Their second argument is sharper and less quotable. They contend that graduate training in evaluating empirical evidence is thin, leaving many early-career clinicians without the tools to distinguish a modality with two trials from one with two hundred, and therefore vulnerable to enthusiastic overapplication. That is a criticism of professional training as much as of IFS, and it deserves to land somewhere other than on the model alone.

Overapplication to conditions nobody has studied

This is the criticism we think is strongest, and it is a safety argument rather than an evidence one.

Brownstone and colleagues name psychosis and severe dissociative conditions specifically. The concern is coherent on its face. A model whose core technique involves deliberately experiencing yourself as composed of distinct parts, each with its own perspective and voice, carries an obvious theoretical risk for someone whose difficulty is already with reality testing or with an unstable sense of a unified self. As critics have put it, encouraging the splitting of the self into parts for people who struggle with reality testing may be disorganising rather than clarifying.

Nobody has run the study that would settle this. Meanwhile IFS is offered widely, including to people with exactly these presentations, on the strength of a model that has never been tested in them. "We have not studied it there" and "it is fine there" are different claims, and the field has not always been careful about which one it is making.

The 2025 press wave

Two mainstream pieces shaped the current conversation. In April 2025, Jess McAllen wrote "The Rapid and Dubious Rise of the Internet's New Favorite Therapy" for The Nation, appearing in the May 2025 issue under the cover line "Trauma's New Look." It documented the IFS Institute's growth from training around 500 providers in 2017 to nearly 4,000 a year, a waiting list of roughly 20,000 prospective therapists, and a training ladder costing several thousand dollars per level. It also noted a distinction the public rarely catches: "IFS Trained" and "IFS Certified" are different statuses, and the more common one is the lesser one.

Later in 2025, New York Magazine and The Cut reported a growing number of psychiatrists and psychologists issuing warnings about IFS, casting it as a simplistic allegory and, at worst, a dangerous pseudoscience. The IFS Institute's response was that the cases cited were extreme misapplications of the model, distinct from what standard training teaches.

That response is not unreasonable, and it is also not sufficient. When a model trains 4,000 providers a year into a market with a 20,000-person waiting list, how it behaves in the hands of the median practitioner is a property of the model's dissemination strategy rather than an unlucky external accident. A field that scales that fast owns its misapplications to a degree that "that is not real IFS" does not cover.

The parts nobody in IFS enjoys discussing

Skipping this section is exactly what makes defences of IFS read as marketing, so here it is.

The single, unbroken Self

IFS holds that every person has a core Self that is undamaged, cannot be destroyed by trauma, and possesses qualities like calm, curiosity and compassion inherently rather than developmentally. Clinically this is a powerful stance, and for many people it is the first thing that has ever made self-directed work feel safe. Empirically it is not a finding. It is a philosophical commitment, closer to a claim about human nature than to a hypothesis anyone has tested, and it should be described that way rather than smuggled in as though research had established it.

Unattached Burdens

This is the genuinely awkward one. Alongside parts, some IFS practice describes Unattached Burdens: entities held to be not native to the person, and called "critters" in the model's earlier vocabulary. Some practitioners treat this as a metaphor for internalised material that has no home elsewhere in the system. Others describe it in openly spiritual terms, up to and including literal demons, with an unburdening process critics have compared to exorcism.

There are two separate problems here. The first is that it is untestable and reputationally radioactive, and the model's own leadership has been visibly cautious about publishing on it. The second is clinical, and it is the one that should worry practitioners more. If a therapist treats what is actually a malignant introject, the internalised voice of someone who hurt you, as a foreign entity to be expelled, they may bypass the entire piece of work that would have mattered. The human origin of the suffering gets obscured at the exact moment it needed to be faced.

None of this is central to the parts framework. A person can find protector parts a genuinely useful way of understanding self-sabotage and reject Unattached Burdens entirely, and many practitioners do. But the concept exists inside the model rather than in a caricature of it, and pretending otherwise is not a defence anyone should accept.

The critique from inside

The sharpest criticism is not from the psychiatrists. It is internal, and it is more interesting.

Narrative therapist Chris Hoff has argued that by casting the self as inherently divided, IFS may mirror and reinforce the fragmentation of contemporary life rather than heal it, and that a relentless focus inward risks converting social and political problems into private psychological ones. You do not have to accept the conclusion to notice the observation underneath it, which is that a therapy fluent in the language of self-optimisation will always be at risk of individualising things that are not individual. Practitioners pushed back hard in his comments, which is itself worth reading.

Running alongside that is what people in the community call the "bro-ification" of IFS: the coach-influencer wing, largely outside any clinical governance, selling parts work as a performance and productivity system. That critique comes from people who like IFS, which is precisely what gives it teeth. The model's popularity created a market that the model's training pipeline does not control.

What is coming

The evidence base is thin, and it is also moving, and the second half of that sentence matters.

The IFS Institute's own research page has notably stopped overclaiming. It now describes the evidence as emerging but limited in breadth and depth, notes that depression is the only condition showing statistically significant improvement across multiple pilot randomised trials, states that the risk of harm and adverse events requires systematic study, and concludes with a sentence that a decade of promotional copy would not have contained: "Larger fully-powered clinical trials are needed for each clinical indication." An institution conceding the limits of its own evidence is a better signal than another testimonial.

Meanwhile the trial pipeline has started to move. A randomised trial of IFS for borderline personality disorder, sponsored by the Waypoint Centre for Mental Health Care, is recruiting now, with a February 2026 start and primary completion estimated for December 2027. It is small, with an estimated enrolment of 15, so it will be a feasibility signal rather than a verdict. That is still how a field builds toward the fully-powered trials it needs.

So where does this leave you

If you are deciding whether to try parts work, the evidence supports a modest and useful conclusion. IFS is a plausible, reasonably safe approach for the ordinary difficulties most people bring to it, with the best support in chronic pain, depression and PTSD, and with real reasons for caution if you live with psychosis or a severe dissociative condition. It is not a proven treatment in the sense that CBT is proven, and if someone tells you it is, they have not read the literature.

If you are a practitioner, the case for care is not that the criticism is unfair. It is that most of the criticism is about reach exceeding evidence, and reach is the one thing practitioners individually control.

The strongest thing anyone can say for IFS right now is that it is a promising model with a thin but real and improving evidence base, considerably smaller than that of CBT or EMDR, applied far more widely than that base licenses, and now beginning to be tested properly. That is a defensible position. It is also, notably, the position the IFS Institute itself has arrived at.

Anything stronger is marketing. Anything weaker is a hit piece. The interesting work is in between, and it is being done now.

If you want to try the ideas without the claims

You do not need to resolve the evidence debate to test whether the framework describes anything true about you. That is a smaller question, and you can answer it yourself.

The most common early stumble is worth knowing about in advance. Many people conclude that the parts they find feel invented, which we wrote about in the IFS "autofill" problem. If you want something structured to work from, our free IFS worksheets cover parts mapping, protector interviews and trigger tracking on paper, and the IFS parts assessment gives you a starting map. If you are weighing tools rather than reading, we compared the available IFS assessment tools. And if you have particular conditions in mind, there are separate pieces on IFS and chronic pain, the area with the clearest trial evidence behind it, and IFS for ADHD.