The confident answer to this question, in both directions, outruns the evidence. “Narcissists never change” and “anyone can change with the right therapist” are both said with more certainty than the research supports. What actually exists is a small, honest, and genuinely informative body of work, and it says something more useful than either slogan.
Our overview of narcissism covers the trait, the disorder, and the difference between them, which matters here: this article is specifically about narcissistic personality disorder, the diagnosis, not about difficult people in general.
What the evidence actually is
Start with what is missing. No form of psychotherapy or medication has been tested for narcissistic personality disorder in a randomised controlled trial. The field’s own recent review states that plainly, [weinberg-2022-progress] and it means nobody can point to trial-grade evidence that any specific treatment works, for anybody.
What exists instead is a case series, a less rigorous but still real form of evidence. [weinberg-2024-case-series] followed eight patients who had been diagnosed with narcissistic personality disorder through 2.5 to 5 years of psychotherapy. The patients selected for the report were the ones who improved, and the improvement was substantial: some no longer met the diagnostic criteria for the disorder by the end of treatment, with better functioning documented alongside it.
That is real change, observed and reported. It is not proof that change is common, only that it is possible, in people who received years of sustained treatment and stayed in it.
Why “they never change” feels true anyway
The same literature explains the gap between what is possible and what usually happens, and the explanation is not mysterious.
A formal diagnosis of narcissistic personality disorder is associated with a 63 to 64 percent dropout rate from psychotherapy. Most people who might eventually change are gone from treatment long before the years accumulate that the case series suggests are needed. The disorder that would benefit from sustained treatment is also, structurally, one that makes sustained treatment unusually hard to stay in: difficulty tolerating the therapist’s authority, shame that surfaces as anger rather than reflection, and a self-image that the early stages of therapy directly threaten.
A schematic built from the dropout and duration figures reported in Weinberg and Ronningstam (2022) and Weinberg et al. (2024). Not a single measured cohort.
Read that as a shape, not a measurement: the numbers are illustrative, built from the dropout rate and the treatment length in the two papers cited here, not a single tracked cohort. What it is meant to show is real. The gap between “can this change” and “does this usually change” is almost entirely explained by how many people leave, and when.
The five years is not incidental
The case series measured change at 2.5 to 5 years. That is not a floor set for convenience; it appears to be roughly how long the process actually takes. No study has reported meaningful change on a shorter timeline, and the mechanism proposed for why it takes this long is specific: the therapy has to survive repeated ruptures, where the patient’s grandiosity or shame collides with what therapy asks of them, and each rupture has to be worked through rather than ended.
That has a direct practical implication. A relationship, a workplace, or a family that is waiting for change on the strength of a few months of therapy, or a single difficult conversation, is not comparing itself to anything the evidence describes.
Is this the shape the evidence describes?
Think of a specific person you are wondering about. Tick what is actually true, not what you hope is true. This produces no diagnosis.
0 of 5 ticked
What you are describing resembles the conditions under which change was actually documented. That is worth taking seriously, while remembering that a case series of eight people is a small basis for any prediction about one more person.
Partial engagement with treatment is common and is not nothing, but the evidence here specifically describes sustained, multi-year work. Fewer than four of these is a different, earlier situation.
Hoping, waiting, or noticing a good week is not what the studies measured. That does not mean nothing is happening, only that it is not the specific process this evidence speaks to.
No screener on this site diagnoses a personality disorder. The hub above covers anxiety, depression, stress, sleep, burnout, self-esteem and loneliness.
What this does not establish
Eight patients is a case series, not a trial, and the patients reported on were the ones who improved. A case series cannot say how many people who start this kind of treatment end up like them, only that reaching that outcome is possible for at least some.
There is no comparison group. Nothing here rules out that these eight people would have changed regardless of the specific treatment, though the years involved make a spontaneous explanation unlikely.
And none of this addresses milder narcissistic traits that fall short of a diagnosis, which are far more common and may behave completely differently, including responding to much shorter interventions or to nothing formal at all. If you are trying to work out whether a pattern in your own life is narcissistic personality disorder or something else, our guide to covert narcissism covers a presentation that is often missed entirely because it does not look grandiose.
What actually helps
If it is you. The evidence says the door is open and describes what walking through it costs: years, a qualified therapist experienced with personality disorders, and staying through the parts that do not feel good. That is a realistic basis for trying, not a promise of the outcome.
If it is someone else. Waiting for change you cannot verify is happening is different from supporting change you can see documented. Ask what is actually happening in their treatment, over what timeframe, rather than reading a good week as proof.
Set your own timeline honestly. If someone is not in the kind of process this evidence describes, the research does not support expecting the outcome it describes either. That is not a verdict on the person; it is a statement about what predicts what.
Protect your own functioning regardless. Whether or not someone else changes, on what timeline, is a separate question from what you need in the meantime, and it is the one you have more control over.
When to seek help
Speak to a mental health professional if you recognise the pattern described here in yourself and want to know whether treatment could help, or if a relationship with someone who may have this diagnosis is affecting your own wellbeing. A personality disorder diagnosis is made by a qualified clinician over more than one conversation, not from a description in an article.
If you are having thoughts of harming yourself, treat that as urgent and contact your local emergency services or a crisis helpline.
How MyFreud can help
MyFreud is useful here for tracking your own experience over the long timeframes this research describes. If you are the one trying to change, logging mood and specific incidents shows you patterns a single hard week cannot. If you are watching someone else, the same log separates a genuinely different month from the story you want to be true.
Download MyFreud and start today: App Store or Google Play.