Mentalization-Based Therapy: How MBT Works and for Whom
Mentalization-based therapy, or MBT, is Anthony Bateman and Peter Fonagy's treatment built around one capacity: holding in mind that behavior, your own and other people's, comes from mental states that can be misread. This guide covers what mentalizing is, the three ways it collapses, what a course of MBT involves, and what the trials in borderline personality disorder actually found.
Reviewed byMichael Callans, MSWKey terms in MBT
Three ways mentalizing collapses
Psychic equivalence
Inner experience is taken as reality with no gap for doubt. 'I feel worthless, so I am worthless.' 'Her tone meant contempt.' Alternatives are not rejected so much as unavailable. The way back is slow, concrete reconstruction of what was actually said and done, not reassurance.
Pretend mode
Thinking detached from feeling. Fluent, insightful-sounding accounts of oneself that go nowhere and change nothing, often with a flat quality a listener can sense. Therapy can collude in this for months. The way back is grounding every statement in a specific recent moment.
Teleological mode
Only physical outcomes count as evidence of a mental state. Care has to be proved by someone dropping plans and arriving; reassurance in words means nothing. It produces escalating tests and demands. The way back is naming the test out loud and looking at what it is trying to find out.
The mentalizing pause: use it after the heat drops
- Stop and name the state of the body first: heart rate, jaw, chest, hands. When these are at full tilt, mentalizing is offline and nothing on this sheet will work yet.
- What am I feeling? One or two words, feelings rather than verdicts: 'humiliated and panicky', not 'he was being a bully'.
- What did I conclude about them, stated as the accusation it is? Write it bluntly: 'she thinks I am incompetent and was enjoying it.'
- What might they have been feeling and thinking? Guess, mark it as a guess, and allow it to be ordinary: tired, distracted, worried about something unrelated, badly briefed.
- What else could explain what they did? Write at least two alternatives, including one that makes neither of you the villain. Note which parts of your conclusion were observed and which were inferred.
- What is the smallest next move? One question you could actually ask them to find out, or one thing you will deliberately not do. Then note what happened when you tried it.
Signs your mentalizing has gone offline
- You are certain what the other person meant, and that certainty arrived before any checking.
- Their motives feel single and simple: they wanted to hurt you, full stop.
- The feeling and the fact have merged: it feels like rejection, so rejection is what occurred.
- You need proof in actions rather than words, and the proof required keeps getting larger.
- You can explain yourself at length and nothing shifts, in the conversation or afterward.
- Your own state is hard to name at all, or comes out only as a physical description.
- The urge to do something irreversible, send it, end it, leave, is strongest right now.
What a course of MBT usually involves
- An introductory psychoeducation course, MBT-I, covering the model before treatment proper starts.
- A written formulation made with you: the patterns, what tends to escalate, and what the plan is when it does.
- Weekly individual therapy, focused on moments where mentalizing dropped rather than on history for its own sake.
- A weekly group, where the capacity gets practiced under live interpersonal pressure. Central to the treatment, not an add-on.
- A total length of roughly twelve to eighteen months in the classic outpatient model.
- Progress judged by relationships and by how quickly you recover the capacity, alongside symptom measures.
- Clinicians trained specifically in MBT and in supervision, since the stance is easy to describe and hard to hold.
What the trials found
Bateman and Fonagy's 1999 trial compared psychoanalytically oriented partial hospitalization with standard psychiatric care and found improvement across measures in the treated group over the 18-month program; their 2001 follow-up paper reported the gains sustained 18 months after treatment ended. Their 2009 trial randomized 134 patients with borderline personality disorder to outpatient MBT or to structured clinical management, a strong comparison condition rather than a placeholder. Both arms improved; MBT showed a steeper decline on the composite primary outcome of self-injurious acts, suicide attempts and hospitalization.
The 2020 Cochrane review is the independent check. Across 75 trials and 4,507 participants, MBT compared with treatment as usual reduced self-harm (risk ratio 0.62, three trials) and suicidality (risk ratio 0.10, three trials), with the certainty of the evidence rated low. The review found no clear evidence that any single type of psychotherapy beat the others on symptom severity or psychosocial functioning, which is the fairest summary available: this is a credible treatment for a serious condition, tested in a small number of trials, several of them run by the people who designed it.
Related tools & tests
Looking for therapy that works on this?
Our directory lists licensed clinicians who work with personality disorders and emotion regulation, with their training on each profile. Free to search.
Therapists Who Work With Personality disorders
A worksheet is a start. These licensed therapists work with the same problem; every member is matched to the federal NPI Registry and board checks are dated on the profile.
- View profile
-
View profile
Adrianna CrouchMS, LCPC, LMHC
License verified, checked Sep 29, 2026
“Pivot Back to You”
Baltimore, Maryland Telehealth $155 to $175 per session
-
View profile
Brian FletcherLCSW
License verified, checked Oct 6, 2026
“Faith-Based Counseling for Healing, Growth, and Lasting Change”
Cumberland, Rhode Island Telehealth and in person $125 to $150 per session
About this tool
Mentalization-based therapy, also called mentalization-based treatment and usually shortened to MBT, is a psychotherapy built on a single capacity: the ability to understand behavior, your own and other people's, in terms of the mental states behind it. Thoughts, feelings, intentions and beliefs cannot be seen, so they always have to be inferred, which means they can always be inferred wrongly. MBT treats the loss of that capacity under emotional pressure as the engine of a lot of relational chaos, and rebuilding it as the treatment. It was developed in the 1990s by the psychiatrist Anthony Bateman and the psychologist Peter Fonagy, originally for people with borderline personality disorder, and grew out of attachment research rather than out of symptom-focused behavioral work.
Mentalizing is more specific than empathy or self-awareness. Fonagy and colleagues describe it along four dimensions: automatic or implicit versus controlled or explicit, attention to self versus attention to others, cognitive content such as beliefs versus affective content such as feelings, and reading internal cues versus reading external behavior. Good mentalizing is a balance across all four. Most of it happens implicitly and fast, as it must in ordinary conversation. The key clinical observation is that it is not a stable trait but a fluctuating capacity: it shrinks when the attachment system is aroused, which is precisely when relationships get difficult and accurate reading matters most.
The developmental account explains why some people lose it faster. On this view the capacity is learned by being accurately mentalized: a caregiver who notices your state, reflects it back in a form that is recognizable but marked as theirs rather than yours, and so hands you a usable map of your own insides. Where that mirroring was absent, frightening or systematically wrong, the map comes out patchy, and feelings stay physical and overwhelming rather than becoming things one can think about. Fonagy and Elizabeth Allison added a related idea in 2014: epistemic trust, the willingness to treat what another person tells you as reliable and worth learning from, which is damaged by the same histories and which therapy has to earn back before anything else works.
When mentalizing fails, it does not fail into nothing. Three recognizable modes take over. In psychic equivalence, inner experience is reality: if it feels like contempt it is contempt, and no alternative reading is even thinkable. In pretend mode, thought floats free of feeling, producing long, articulate, oddly dead accounts of oneself that change nothing, something therapy itself can become complicit in. In teleological mode, only physical outcomes count as proof of a mental state: you do not believe someone cares unless they drop everything and arrive, which sets up demands nobody can keep meeting. Naming which mode is running is often the most useful thing a session does.
Read the diagram as text
Two matched scenes, each with two heads facing each other. In both, a dashed thought bubble sits above the second head: that person's actual state, never directly visible. The scenes differ only in the first head's bubble. On the left it is a dashed brand outline holding a small sketch of the other person and a feeling, a reading marked as a guess, and a brand path curves back from the other head to the first: it was checked. On the right the identical bubble is a solid accent shape with nothing inside, the reading taken as reality, and the only path is the neutral one running one way. The accent marks the inference about the other mind. On the phone canvas the two scenes stack.
Holding the other person's mind in mind
Conceptual contrast, not ratings
Thoughts, feelings and intentions cannot be seen, so they always have to be inferred, which means they can always be inferred wrongly.
Mentalizing holds: What the other person might be thinking is held as one reading among others, and checked with them; the therapist's not-knowing stance models that move rather than demonstrating insight.
Mentalizing collapses into psychic equivalence: The inner reading becomes reality: if it feels like contempt it is contempt, no alternative is thinkable, and nothing is checked.
The therapist's posture has a name: the not-knowing stance, sometimes called the inquisitive stance. The clinician deliberately does not present as the one who can see into you. They ask, they wonder aloud, they check, they say when they are confused, and they are willing to be corrected, because the point is to model the activity rather than demonstrate expertise in it. When mentalizing collapses mid-session, the standard move is to stop and rewind: pause the content, go back to just before the moment the temperature changed, and work out slowly what each person was thinking and feeling there. MBT deliberately avoids the deep interpretations of unconscious meaning associated with older psychoanalytic practice, on the grounds that they ask for a capacity the person has temporarily lost.
A full course is substantial and structured. Classic outpatient MBT runs twelve to eighteen months and combines weekly individual sessions with a weekly group, which is not an optional extra: the group is where mentalizing gets practiced under live interpersonal pressure. Most programs start with an introductory psychoeducation course, MBT-I, explaining the model itself, then write a shared formulation naming the patterns and the risks and a plan for what to do when things escalate. Progress is judged by relational capacity as much as by symptoms. A version adapted for younger people, MBT-A, also exists; this page describes the adult treatment.
The evidence began with two trials by the developers. In 1999 they reported a randomized comparison of psychoanalytically oriented partial hospitalization against standard psychiatric care, with the treated group improving across measures over the 18-month treatment while the control group largely did not; a separate 2001 paper reported the gains holding at an 18-month follow-up after treatment ended. The stronger test came in 2009: 134 patients with borderline personality disorder randomized to outpatient MBT or to structured clinical management, which was itself a well-organized protocol rather than a token comparison. Both groups improved substantially; the MBT group declined more steeply on the primary outcome, a composite of self-injurious acts, suicide attempts and hospitalization, and on self-reported problems.
The independent summary is more measured, and worth reading before anyone calls MBT proven. The 2020 Cochrane review of psychological therapies for borderline personality disorder covered 75 randomized trials and 4,507 participants. Comparing MBT with treatment as usual, it found reduced self-harm (risk ratio 0.62 across three trials) and reduced suicidality (risk ratio 0.10 across three trials), with a depression result whose confidence interval crossed zero, and rated the certainty of that evidence low, meaning further trials could change the picture. The same review found no clear evidence that any one type of psychotherapy outperformed the others for symptom severity or psychosocial functioning. MBT has good support for a serious condition and a thinner trial base than its reputation implies.
Beyond borderline personality disorder, MBT has been adapted for antisocial personality disorder, eating disorders, substance use, depression and family work, and the evidence in those areas is considerably weaker, mostly small studies and uncontrolled series. The practical limits matter too. Full programs exist mainly in specialist public services and a handful of private settings, so availability is the common obstacle; the group component rules it out for some people; and a twelve to eighteen month commitment is not a small ask. Dialectical behavior therapy has more trials behind it for this population, and the Cochrane review found no clear winner between them, so availability, fit and whether you can tolerate group work often decide it.
Three things MBT is not. It is not mindfulness: the object of attention is mental states and their effects on behavior, not the breath or present-moment experience as such. It is not training in mind-reading, and a therapist who claims to know what you are feeling is doing the opposite of the model, which treats every reading as a hypothesis to be checked. And it is not insight-hunting; long, clever explanations of oneself are exactly what the model calls pretend mode. The printable below is a single-page version of the central move: stop, locate your own state, consider what the other person's might be, generate one alternative explanation, then pick the smallest next step. It sits well alongside Perspective Taking for the other-person half and Check the Facts Worksheet for testing the interpretation itself.
References
- Bateman A, Fonagy P. Effectiveness of partial hospitalization in the treatment of borderline personality disorder: a randomized controlled trial. American Journal of Psychiatry. 1999;156(10):1563-1569.
- Bateman A, Fonagy P. Randomized controlled trial of outpatient mentalization-based treatment versus structured clinical management for borderline personality disorder. American Journal of Psychiatry. 2009;166(12):1355-1364.
- Bateman A, Fonagy P. Mentalization-Based Treatment for Personality Disorders: A Practical Guide. Oxford University Press; 2016.
- Fonagy P, Luyten P. A developmental, mentalization-based approach to the understanding and treatment of borderline personality disorder. Development and Psychopathology. 2009;21(4):1355-1381.
- Fonagy P, Allison E. The role of mentalizing and epistemic trust in the therapeutic relationship. Psychotherapy. 2014;51(3):372-380.
- Storebø OJ, Stoffers-Winterling JM, Völlm BA, et al. Psychological therapies for people with borderline personality disorder. Cochrane Database of Systematic Reviews. 2020;(5):CD012955. doi:10.1002/14651858.CD012955.pub2
- Bateman A, Fonagy P. Treatment of borderline personality disorder with psychoanalytically oriented partial hospitalization: an 18-month follow-up. Am J Psychiatry. 2001;158(1):36-42. https://doi.org/10.1176/appi.ajp.158.1.36
Cite this source
Psychology.com. (2026, November 6). Mentalization-Based Therapy (MBT). Psychology.com. https://psychology.com/tools/mentalization-based-therapy
Mentalization-Based Therapy (MBT) FAQ
What is mentalization?
Mentalization, or mentalizing, is the capacity to understand behavior in terms of mental states: the thoughts, feelings, intentions and beliefs behind what people do, including what you do yourself. Because none of those can be observed directly, mentalizing is always inference, which is why it can be wrong. Researchers describe it along four dimensions: fast and implicit versus deliberate and explicit, focused on self versus on others, concerned with thoughts versus with feelings, and reading internal states versus outward behavior. It is a fluctuating capacity rather than a fixed trait, and it reliably weakens when emotion runs high in close relationships.
What is mentalization-based therapy?
Mentalization-based therapy, or MBT, is a treatment developed by Anthony Bateman and Peter Fonagy that works directly on the capacity to mentalize rather than on symptoms or on thought content. The therapist adopts a not-knowing stance, asking and checking rather than interpreting, and when the conversation heats up they stop and rewind to reconstruct what each person was thinking and feeling just before the shift. A full outpatient course usually combines weekly individual sessions with a weekly group over twelve to eighteen months. It was built for borderline personality disorder and has since been adapted for other conditions with less evidence.
What happens in an MBT session?
Less archeology and more slow motion than people expect. Sessions usually start from something recent and specific, an argument, a text message, a moment in last week's group, and work through it a beat at a time: what you felt, what you concluded the other person meant, what else it could have meant, what you did next. The therapist asks more than they explain, says when they are confused, and invites correction. When you lose the thread mid-session, which is the point at which the work becomes real, they pause and go back rather than push forward. Groups do the same thing with more people in the room.
Does MBT work for borderline personality disorder?
The evidence is genuinely supportive and genuinely limited. Bateman and Fonagy's 2009 trial randomized 134 patients to MBT or to structured clinical management and found both groups improved, with MBT declining more steeply on a composite of self-injury, suicide attempts and hospitalization. The 2020 Cochrane review of 75 trials reported that MBT reduced self-harm and suicidality against treatment as usual, across three trials for each outcome, but rated the certainty of that evidence low. The same review found no clear difference between the various psychotherapies for this condition. Several of the key trials were run by the treatment's own developers.
MBT or DBT: which is better for me?
No trial has established a winner. The 2020 Cochrane review found no clear evidence that any one psychotherapy outperformed the others for borderline personality disorder, though dialectical behavior therapy has been tested in more trials. The practical differences are clearer than the outcome data. Dialectical behavior therapy teaches specific skills, assigns homework and tracks behavior on diary cards, which suits people who want structure and tools. MBT spends its time on understanding states of mind in particular moments and has less homework. Availability usually settles it, since full programs of either are concentrated in specialist services.
How long does MBT take?
The classic outpatient model runs twelve to eighteen months, with weekly individual sessions plus a weekly group, often preceded by a short introductory course on the model itself. That length is deliberate: the treatment aims at a capacity that fluctuates under pressure, and the group exists so that the pressure happens in a room where it can be worked with. Shorter adaptations exist in day services and inpatient settings, and the original 1999 trial used a partial hospitalization program lasting up to eighteen months. Anyone offering MBT as a brief intervention is describing something different from the tested model.
How do I find an MBT therapist?
Availability is the main obstacle, and it is worth knowing that before you start looking. Complete programs, individual plus group with trained clinicians and supervision, sit mostly inside specialist public services and a small number of private clinics, so a referral route often matters more than a search. Many therapists in general practice have MBT training and use the stance in individual work without the group, which is a reasonable thing to ask about directly: which training they completed, whether they offer the full model or an individual adaptation, and what they do when someone is in acute difficulty between sessions.