Can Personality Disorders Improve? Long-Term Remission, Recovery and Relapse

Bottom line: Personality disorders are not an unchangeable identity sentence. Long-term studies repeatedly find that symptoms and categorical diagnoses can change substantially, even when some habitual traits and functional difficulties persist. In a 2019 meta-analysis of 11 prospective studies and 837 adults with BPD, approximately 50–70% achieved remission over follow-up. That finding is hopeful, but it is not a forecast for any individual and not proof that a specific treatment caused those remissions. Recovery means more than becoming symptom-free. [1–4,7]

Research question and evidence approach

Question: How often do adults diagnosed with a personality disorder improve over time, how should remission be distinguished from recovery, and what can longitudinal and randomized studies actually tell us about prognosis?

We prioritized prospectively observed clinical populations over anecdotal before-and-after stories. A 2019 published BPD meta-analysis anchors the quantitative discussion; longstanding cohort reviews, a ten-year dimensional-versus-categorical prediction study, the 2024 BPD treatment guideline, randomized schema-therapy evidence and 2025 work on psychosocial recovery add distinct context. [1–8]

This is a targeted, critical review of published evidence, not an original or updated quantitative meta-analysis. There is no newly performed literature-wide search with prespecified inclusion criteria, blinded duplicate study selection, original participant-level pooling, or de novo forest plot. Published percentages and sample sizes are credited to the original study. We refuse to combine people with different personality-disorder diagnoses into a single unsupported “recovery rate.” [1–4]

What does 'enduring' mean, clinically?

Diagnostic manuals describe personality disorders using relatively persistent patterns of difficulties in self and interpersonal functioning. That does not mean a person's whole personality is fixed forever, that relationships cannot improve, or that one diagnosis makes meaningful change impossible. The apparent stability of a personality-disorder diagnosis depends partly on its criteria, the clinician's measurements and the length of follow-up. [2–4]

A personality pattern can be longstanding and responsive to new environments, greater safety, learning, psychological therapy, substance-use treatment, medication for co-occurring disorders and changing life responsibilities. Likewise, a familiar trait may persist after a diagnosis is no longer met. For example, someone may remain temperamentally sensitive to criticism but learn to manage conflict, pursue relationships, sustain work and express needs clearly. Change need not erase individuality. [2,3,7]

Three different outcomes often confused in headlines

These are complementary but not identical. A person may report fewer crises yet remain isolated and unable to work; another may retain measurable symptoms while building a satisfying life. A meaningful research article should ask which endpoint a paper measured before quoting its success rate. [1,2,7,8]

The strongest numerical anchor: the 2019 BPD meta-analysis

Álvarez-Tomás and colleagues reviewed prospective adult BPD studies with at least five years' follow-up. Their systematic search covered several psychological and medical databases and identified 11 studies involving 837 participants from nine countries. The studies followed people for approximately five to fourteen years, with a median follow-up of six years. Four outcomes were eligible for meta-analytic synthesis: diagnostic remission, completed suicide, depressive symptoms and functioning. [1]

The reported long-term proportion achieving BPD diagnostic remission was around 50–70%, depending on the included studies and follow-up approach. Depressive symptoms and functional impairment also tended to improve, but functional gains were more limited than symptom remission might imply. The review reported completed-suicide estimates in the 2–5% range across included cohorts—a serious outcome that reinforces the need for appropriate care and careful, non-sensational reporting. [1]

The remission range is not the probability that a newly diagnosed individual will be well in exactly five or ten years. The samples were drawn from clinical settings; people differed in baseline severity, comorbidities, treatments, recruitment setting, social circumstances and how remission was defined. Some studies experienced participant attrition, and estimates at a chosen follow-up differ from the proportion who ever experienced remission. [1]

Most importantly, this synthesis does not establish the causal contribution of dialectical behavior therapy, schema therapy, spontaneous change, supportive relationships or any one pathway. Naturalistic follow-ups include overlapping exposures, and the review could not draw robust conclusions about the long-term effects of specific treatments from the moderators examined. [1]

Why an improved diagnosis is not the same as restored functioning

The major longitudinal reviews highlight a recurring pattern: personality-disorder symptom profiles can change substantially, but interpersonal and occupational problems may linger. People may find their emotions less overwhelming while remaining behind in education, finances, career development, social connection or confidence after years of illness and isolation. That gap between symptom remission and fuller recovery is not a moral weakness; it is simply a different outcome domain. [2,3,7]

A 2025 clinical review of psychosocial functioning emphasizes why treatment systems should measure function directly. An improvement in a checklist score is important, but a person may care more about making a friend, keeping an apartment, rebuilding trust or pursuing an educational goal. These outcomes require longer observation and sometimes practical social interventions beyond traditional symptom-focused psychotherapy. [7]

Recovery should therefore be evaluated longitudinally and collaboratively. An assessment might ask how often someone can identify and manage emotions, whether they can navigate disagreements without a crisis, whether they experience satisfying connection, and whether they can pursue goals that matter to them. Those are indicators of life functioning, not a secret list of traits a stranger must exhibit to be considered lovable. [7,8]

What ten-year studies tell us about traits and diagnoses

Skodol's 2008 review of major prospective studies reported unexpectedly substantial reductions in categorical personality pathology over time, alongside persistent vulnerabilities in functioning for some people. Later work with colleagues also found that maladaptive traits could remain more stable than categorical diagnostic status. [2,3]

A ten-year follow-up from the Collaborative Longitudinal Personality Disorders Study compared the predictive performance of different diagnostic and trait models. Morey and colleagues reported that dimensional symptom counts and trait information often predicted future outcomes better than treating a person's named disorder as an all-or-nothing category. This does not invalidate diagnoses; it means the degree and kind of ongoing difficulty can be more informative than a single label alone. [4]

From a patient's point of view, the distinction can be liberating. No longer crossing a threshold is good news. Yet the remaining traits may still deserve care and accommodations, and a clinician should measure changes in actual functioning rather than declare a person “fixed.” From a researcher's point of view, estimates based on different diagnostic generations cannot automatically be pooled. [2–4]

Is the prognosis the same for all ten named personality disorders?

No. The best-characterized long-term data in this review center on BPD and several mixed cohorts. The 50–70% BPD remission range must not be generalized to avoidant, schizoid, paranoid, narcissistic, dependent, histrionic, antisocial or obsessive-compulsive personality disorders. Differences in severity, diagnoses, service access and measurement mean that a single cross-diagnosis prognosis percentage would be misleading. [1–4]

Research on people with avoidant personality disorder, for example, can examine social anxiety and personality manifestations separately. Some treatment trials and naturalistic cohorts include AVPD alongside other disorders, but direct multi-decade data and matched comparisons remain much more limited for some diagnoses. The absence of precise estimates does not show that a condition is permanent or untreatable; it marks an evidence gap. [2,3,6]

ICD-11's dimensional emphasis on severity and trait profiles may help clinicians describe change more continuously than a simple yes/no categorical threshold. Such models still need prospective validation and should not be used to erase patient experiences or practical support needs. [4,7]

What evidence can treatment studies add?

Prognosis cohorts tell us what happened over time, while randomized trials attempt to estimate what changed because of an assigned intervention compared with another condition. Both are important but answer different questions. A patient who improves over ten years might have received several therapies or none, moved somewhere safer, changed jobs, built a support system or received care for depression and substance use. A naturalistic cohort cannot isolate the effect of one component without strong assumptions. [1,5,6]

For BPD, the 2024 American Psychiatric Association guideline emphasizes structured, evidence-based psychotherapy and careful clinical management. For several personality disorders, Bamelis and colleagues' randomized study of 323 patients found schema therapy associated with higher rates of personality-disorder recovery than its comparisons at three-year follow-up. However, the trial included multiple diagnoses and a particular outpatient service context. That result does not supply a general ten-year remission rate or prove schema therapy is superior for every personality disorder. [5,6]

A clinical intervention may improve emotion regulation without immediately changing relationship satisfaction. A comparison using recovery from a named diagnosis cannot be treated as if it measured suicide prevention, quality of life and sustained financial independence. Real scientific precision starts by protecting those boundaries. [5,6]

What does recovery mean to people living with a diagnosis?

A 2025 mixed-methods concept-mapping study examined views from patients, significant others and therapists. Its participants identified recovery themes including understanding and recognizing oneself, regulating emotion, directing one's life, and connecting with others. This is not an experimental estimate of what causes recovery, nor a representative population survey. It offers a useful reminder that a diagnostic questionnaire may miss what people value most. [8]

The emphasis on self-understanding also counters two stereotypes. One is that anyone who has ever met criteria for a personality disorder cannot grow or repair a relationship. The other is that improvement requires becoming an entirely different person. Recovery can involve learning to notice early warning signs, accepting responsibility for behavior, treating co-occurring conditions, communicating more clearly and making choices aligned with values. [7,8]

Relapse, recurrence and uneven progress

Remission studies differ in how often people are evaluated and what counts as recurrence. A person may experience one renewed period of intense distress without meeting the full diagnostic criteria again. Conversely, someone may no longer qualify for a personality disorder yet continue to struggle with depression, addiction, suicidality or economic instability. A long-term plan should address these separately rather than treating any setback as proof recovery was never real. [1–3]

Time and environment matter. Recovery may accelerate or stall around bereavement, violence, social isolation, unstable housing or poor access to care. Those are plausible influences on real-world functioning, but this literature does not justify assigning a universal “recovery timeline” or blaming a patient for slow progress. [1,7]

Important caveats: scientific and ethical

• Selection bias: Clinical cohorts can overrepresent people with greater severity and treatment contact; people lost to follow-up may differ from those retained. [1]
• Historical diagnostic changes: DSM categorical criteria and ICD-11 dimensional formulations are not directly interchangeable, so estimates are not always comparable across generations. [2–4]
• No guaranteed treatment effect: Published long-term remission is not a randomized trial of one therapy against no care. [1,5,6]
• Functional recovery needs measurement: A person can stop meeting full criteria while remaining at high social or economic disadvantage. [7]
• No diagnosis-by-anecdote: A partner's behavior, a breakup, one angry message or another person's claim about someone's personality cannot establish any diagnosis or prognosis. [3,7]

The literature supports careful hope, not false assurance. For people experiencing self-harm thoughts, severe impairment or unsafe relationships, timely clinical and safety support matters regardless of whether a diagnostic label is stable. [5,7]

Evidence-based verdict

Personality disorders are not necessarily lifelong in their full diagnostic form. The best available longitudinal studies show substantive improvements in many people, with BPD follow-ups yielding remission in roughly half to two-thirds of studied adults over years. Nevertheless, residual functioning challenges are real, the evidence base is uneven across diagnoses, and recovery definitions matter enormously. [1–4]

A good treatment plan should monitor symptoms, relationships, school/work, quality of life and personal priorities; acknowledge uncertainty; and make room for progress that is slow, non-linear and meaningful. The evidence does not support labeling anyone beyond hope, nor does it permit predicting exactly how a particular person will change. [5–8]

Related THS research: BPD relationships and evidence-based care, BPD versus complex PTSD, AVPD versus social anxiety, and the mental-health guides hub.