F60.3: A Therapist's Reference for Borderline Personality Disorder

Borderline personality disorder is one of the most clinically demanding diagnoses in outpatient practice, and it is also one of the most miscoded. The condition sits at the intersection of affect regulation, identity, attachment, and impulsivity, which means the chart that supports it has to do more than list a label. It has to show the personality-level pattern, count the criteria met, document the suicide risk that travels with the diagnosis, and rule out the conditions that look like BPD on the surface but call for a different treatment plan.

The code is also a name worth getting right. The US ICD-10-CM descriptor for F60.3 is “Borderline personality disorder”, which matches the DSM-5 name. The original WHO ICD-10 uses a different descriptor, “Emotionally unstable personality disorder, borderline type”, for the same clinical picture. Charts under US payer contracts should use the ICD-10-CM descriptor on the diagnostic line; therapists working with international clients or under non-US frameworks may encounter the WHO ICD-10 wording in older literature and templates.

For licensed therapists, psychologists, counselors, and clinical social workers who diagnose and bill under ICD-10-CM, this page is part of the ICD-10 codes for therapists sub-hub. The DSM-5 criteria, the differential from bipolar II and complex PTSD, the audit-ready documentation pattern, the personality-level modalities indicated, and the CPT pairings each get their own section below.

Educational reference for licensed mental health practitioners. Coding and documentation requirements vary by state, payer, and setting; verify against your state licensing board, payer contracts, and the current ICD-10-CM official guidelines for the year of service.

What the f60.3 code means

ICD-10-CM defines the code as “Borderline personality disorder” (aligning with the DSM-5 name). It sits inside F60-F69 (Disorders of adult personality and behavior). F60 is the category for “Specific personality disorders,” and the fourth-character specifier names the personality type: F60.0 (paranoid), F60.1 (schizoid), F60.2 (antisocial), the borderline-type code at F60.3, F60.4 (histrionic), F60.5 (anankastic, equivalent to DSM-5 obsessive-compulsive personality disorder), F60.6 (anxious, equivalent to DSM-5 avoidant), F60.7 (dependent), F60.81 (narcissistic), F60.89 (other), and F60.9 (unspecified). The WHO ICD-10 source uses a different descriptor (“Emotionally unstable personality disorder, borderline type”) for the same clinical picture; US therapists code against ICD-10-CM rather than WHO ICD-10.

The WHO ICD-10 system inherited its naming from earlier European nosology, where the construct was framed as “emotionally unstable personality.” WHO ICD-10 carved that construct into two subtypes: an impulsive type and a borderline type, with the borderline-type specifier mapping to the DSM-5 borderline personality disorder diagnosis. When CMS adapted ICD-10 into ICD-10-CM for US billing, the F60.3 descriptor was aligned with the DSM name, so US claims and audit-ready documentation use “Borderline personality disorder” on the diagnostic line with code F60.3. DSM-5 itself uses the numeric code 301.83 in its own coding system, but US payers read against the ICD-10-CM mapping.

The official descriptor is published by the Centers for Medicare and Medicaid Services and mirrored at icd10data.com, with the broader CMS reference available at the CMS ICD-10 page. Use the descriptor wording as published; reviewers expect the specifier language to match the official text rather than a paraphrase, even when the chart also carries the DSM name.

DSM-5 criteria for borderline personality disorder

The DSM-5 criteria for BPD are nine items, five of which must be met to support the diagnosis. The pervasive pattern of instability across interpersonal relationships, self-image, and affect, plus marked impulsivity, must begin by early adulthood and be present across a range of contexts. The summary table below is intended for quick chart-side reference; the full descriptors live in the DSM-5 itself.

Criterion	What it requires	Threshold for the chart
1: Abandonment	Frantic efforts to avoid real or imagined abandonment.	Concrete example of the pattern, not isolated incidents.
2: Unstable relationships	A pattern of unstable and intense interpersonal relationships, characterised by alternating between idealisation and devaluation.	Documented across multiple relationships, not a single conflict.
3: Identity disturbance	Markedly and persistently unstable self-image or sense of self.	Client report of shifting values, goals, vocational direction, or sexual identity.
4: Impulsivity	Impulsivity in at least two areas that are potentially self-damaging (spending, sex, substance use, reckless driving, binge eating).	Two distinct domains, not a single behaviour. Self-injurious or suicidal behaviour is captured separately under criterion 5.
5: Suicidal/self-harm	Recurrent suicidal behaviour, gestures, threats, or self-mutilating behaviour.	Recurrent rather than isolated; document frequency and pattern.
6: Affective instability	Affective instability due to a marked reactivity of mood (intense episodic dysphoria, irritability, or anxiety usually lasting a few hours and rarely more than a few days).	Reactivity is the key word: mood shifts in response to interpersonal triggers, within hours.
7: Emptiness	Chronic feelings of emptiness.	Client’s own language preferred.
8: Anger	Inappropriate, intense anger or difficulty controlling anger (frequent displays of temper, recurrent physical fights).	Observable in or out of session, with concrete examples.
9: Paranoia/dissociation	Transient, stress-related paranoid ideation or severe dissociative symptoms.	Transient is the key word; sustained paranoid or dissociative pictures point toward differential diagnoses.

A defensible chart names each of the five (or more) criteria that apply, with one concrete client report or in-session observation per criterion. Listing the diagnosis without the criterion-by-criterion mapping is the chart pattern that gets flagged on review, because it does not show how the threshold was reached.

When the borderline-type code is the right pick

F60.3 applies when the chart establishes a pervasive pattern across interpersonal, self-image, affect, and impulsivity domains, that pattern began by early adulthood, and at least five of the nine criteria are met with documented evidence. The clinical reality is that BPD presentations sit on a spectrum, and a client can show three or four traits without crossing the threshold. A chart that codes F60.3 with only three or four documented criteria is one of the most common audit failures in personality-disorder coding.

When the picture does not yet meet the full criteria, more general codes are available. F60.9 (unspecified personality disorder) holds the diagnosis open when a personality-level pattern is clear but the specific subtype is still under assessment. F60.89 (other specific personality disorders) is appropriate when traits are mixed across subtypes and no single F60.x code captures the picture. Neither is a long-term substitute for the borderline-type code; both should resolve to a specific code as the assessment matures, and an auditor reading the chart will expect to see that movement.

The differential diagnoses that most frequently overlap with BPD deserve their own paragraph because confusing them is the second most common audit failure after under-counting criteria. The most clinically important differentials are bipolar II disorder, complex PTSD, narcissistic personality disorder, and major depressive disorder with mood reactivity.

Differential	Distinguishing feature
Bipolar II (F31.81)	Bipolar mood shifts are autonomous and last days to weeks; BPD affective instability is reactive to interpersonal triggers and shifts within hours. Family history of bipolar illness, response to mood stabilisers, and clearly demarcated hypomanic episodes favour bipolar II. Reactivity to perceived rejection and rapid same-day mood shifts favour BPD.
Complex PTSD / F43.12 PTSD	Trauma-driven dysregulation can mimic BPD’s affective and interpersonal picture. The differentiator is index trauma exposure, intrusion symptoms, and avoidance behaviour that fit the PTSD architecture. Many clients meet criteria for both; comorbidity is the rule rather than the exception. See F43.10 unspecified PTSD for the trauma-side documentation pattern.
Narcissistic personality disorder (F60.81)	NPD’s grandiosity and entitlement contrast with BPD’s identity instability and emptiness. Overlap is real, and clients can meet criteria for both.
Major depressive disorder with anxious distress	MDD’s mood disturbance is sustained rather than reactive; BPD’s affective instability shifts within hours and is interpersonally driven.

A defensible chart names the differentials considered and the data that ruled them out. “Bipolar II ruled out based on absence of hypomanic episodes, family history, and reactivity-of-mood pattern” is the kind of sentence that holds up under review. “BPD ruled in” without the differential work shown is not.

Suicide risk and the BPD chart

BPD carries a clinically significant elevated suicide risk: estimates from longitudinal studies place lifetime suicide rates in BPD samples between 5% and 10%, with self-injurious behaviour considerably more common. Risk assessment is therefore not an episodic question for the borderline-type chart; it is a standing item on each progress note, with the cadence of reassessment tied to the chronic-but-fluctuating risk profile.

The chart that survives review on this point does five things. First, the intake establishes baseline risk with a structured framework that names ideation, intent, plan, access, protective factors, and history of attempts or self-injury. Second, each progress note records a brief reassessment, with the date and the result, even when nothing has changed. Third, when risk rises, the note documents the safety plan that was reviewed or revised, including means-restriction discussion when access is a concern. Fourth, when self-injurious behaviour or a suicidal gesture occurs between sessions, the next note documents the assessment, the safety plan response, and any consultation or escalation. Fifth, the treatment plan is paired with a modality that has documented effects on suicide risk in BPD samples, which is one of the reasons dialectical behaviour therapy is the most-cited first-line approach.

The American Foundation for Suicide Prevention’s research summary on BPD and suicide and the Linehan Institute’s primary literature on DBT for chronically suicidal clients provide the evidence base most therapists reference when defending the modality choice in chart review. Risk assessment frameworks that pair cleanly with BPD presentations include the Columbia Protocol, which is published and freely available, and the Linehan Risk Assessment and Management Protocol used inside structured DBT programs.

Modalities indicated for borderline personality disorder

The treatment plan attached to a BPD chart should name a personality-level modality rather than a symptom-only approach. Four modalities have the most established evidence base; a fifth, general psychiatric management, has emerged as a reasonable comparator for clients who do not have access to the structured programs.

Dialectical behaviour therapy (DBT), developed by Marsha Linehan, is the most-cited first-line treatment for BPD with active suicidal or self-injurious behaviour. The standard DBT package combines weekly individual therapy, weekly skills group, between-session phone coaching, and a therapist consultation team. The skills modules cover mindfulness, distress tolerance, emotion regulation, and interpersonal effectiveness. For therapists building a DBT-informed practice, the DBT diary card template covers the session-side tracking tool most often referenced in progress notes.

Mentalization-based therapy (MBT), developed by Bateman and Fonagy, focuses on the client’s capacity to think about their own and others’ mental states. MBT is typically delivered as a combination of individual and group sessions over twelve to eighteen months, and has randomised-controlled-trial support comparable to DBT in several head-to-head studies.

Transference-focused psychotherapy (TFP), developed by Kernberg and colleagues, is a psychodynamic treatment that targets the integration of split internal representations through work in the transference. TFP is delivered as twice-weekly individual therapy with a fidelity-monitored treatment manual.

Schema therapy, developed by Young, integrates cognitive, behavioural, and experiential techniques to identify and modify early maladaptive schemas. The modality has growing evidence in BPD samples, with randomised trials showing comparable outcomes to other personality-level approaches.

Good psychiatric management (GPM) was developed by Gunderson as a generalist outpatient framework for clinicians who do not have access to structured DBT or MBT programs. GPM emphasises psychoeducation, case management, suicide-risk monitoring, and a clear treatment contract. The evidence base shows GPM produces outcomes comparable to structured DBT for many clients, which makes it the realistic default for solo-practice or small-group settings.

The choice between these modalities is a function of clinician training, client preference, and the local availability of structured programs. The chart that survives review names the modality, references the evidence base briefly, and documents the treatment frame: session frequency, expected duration, between-session structure, and the suicide-risk monitoring cadence.

Adjacent codes in the F60 family

The F60 family contains all of the specific personality disorders, and clients with BPD often meet criteria for additional personality-level diagnoses. The codes most frequently appearing alongside the borderline-type code on the same chart are summarised below.

Code	Condition	Relevance to BPD
F60.0	Paranoid personality disorder	The transient paranoid ideation in BPD criterion 9 does not by itself warrant an F60.0 add. Pervasive, sustained paranoia points toward this code.
F60.2	Antisocial personality disorder	Comorbidity is real, particularly in men with BPD presentations. The chart should establish the distinct antisocial criteria rather than collapsing the picture into a single label.
F60.6	Anxious (avoidant) personality disorder	Avoidant and borderline patterns share interpersonal hypersensitivity, but the underlying mechanism differs: avoidance vs reactivity.
F60.81	Narcissistic personality disorder	Overlap is clinically common; the chart benefits from naming both codes when both sets of criteria are met.
F60.9	Personality disorder, unspecified	Holds the diagnosis open when a personality-level pattern is clear but the specific subtype is still under assessment. Not a long-term substitute for the borderline-type code.

Comorbid mood, anxiety, trauma, and substance-use codes also commonly appear on the same claim line as BPD. F33.x (major depressive disorder, recurrent) and F32.x (single episode) frequently sit alongside the personality diagnosis. F41.1 (generalized anxiety disorder) and F43.12 (chronic PTSD) appear when the symptom picture also fits those criteria. The F10-F19 substance-use family is also clinically common in BPD caseloads, and the substance comorbidity feeds directly into the impulsivity criterion as well as the suicide risk assessment, because intoxication is a known acute risk multiplier. When the chart reflects two or more distinct disorders, all relevant ICD-10-CM codes can be carried; the diagnostic order on the claim line typically lists the focus-of-treatment code first, with comorbidities as additional pointers.

Documentation that holds up under audit

A defensible BPD chart establishes seven elements at intake and revisits them across the treatment episode.

1. Pervasive pattern documented. The chart names the pervasive pattern across interpersonal relationships, self-image, affect, and impulsivity, with onset by early adulthood and presence across a range of contexts. Single-domain or single-relationship instability does not satisfy the construct.
2. At least five of nine criteria mapped to concrete evidence. Each of the five (or more) criteria is paired with a specific client report or in-session observation. A sample Assessment paragraph reads: “Client meets criteria 1 (frantic abandonment efforts following partner withdrawal last month), 2 (idealisation-devaluation pattern across three close relationships in past year), 4 (impulsivity in spending and substance use), 5 (two suicidal gestures in past six months, ongoing self-injurious cutting), 6 (mood shifts within hours triggered by perceived rejection), and 7 (chronic emptiness in client’s own words).” A chart that lists the diagnosis without mapping the criteria is the chart an auditor flags first.
3. Differential diagnoses ruled out. Bipolar II, complex PTSD, narcissistic personality disorder, and major depression with mood reactivity are each named, with the data that supported ruling each in or out. A line that says “Bipolar II ruled out based on absence of hypomanic episodes and reactivity-of-mood pattern within hours” carries more weight than a blanket “BPD ruled in.”
4. Onset by early adulthood established. Personality disorder criteria require the pattern to have been present across a long horizon. A developmental history that places the pattern in late adolescence or early adulthood, with continuity through the present, supports the diagnosis. A pattern that emerged in midlife in response to a specific stressor more likely fits a stress-reaction or adjustment-disorder picture.
5. Suicide risk assessed at every contact. Each progress note records a brief reassessment, with the date and the result. When risk rises, the note documents the safety plan reviewed or revised, including means-restriction discussion when access is a concern. This is the single most-scrutinised element of a BPD chart on review.
6. Functional impairment described concretely: employment instability, relationship turnover, financial consequences of impulsivity, social isolation following devaluation episodes, missed work days following affective dysregulation. Generic language (“functioning impaired”) is weaker than specific, observable impacts.
7. Treatment plan tied to a personality-level modality. The plan names DBT, MBT, TFP, schema therapy, GPM, or a documented equivalent, with the treatment frame: session frequency, expected duration, between-session structure, and the suicide-risk monitoring cadence. A treatment plan that addresses only the presenting symptom of the week, without a personality-level frame, is the second most common audit failure on BPD charts.

The American Psychological Association’s record-keeping guidance and the HIPAA provisions at 45 CFR § 164.501 distinguish between psychotherapy notes, which are kept separate and receive heightened protection, and the progress notes that go into the medical record. Diagnostic justification belongs in the progress note and the treatment plan, not in psychotherapy notes, because the progress note is what supports the billed diagnosis under audit.

A note on re-documenting at intake and review intervals: personality disorder codes carry a particular audit vulnerability when they are added to a problem list and never revisited. A chart that carries the diagnosis forward through years of treatment without re-documenting the criteria at the annual review, or at any change in treatment plan, looks to a reviewer like a diagnosis that was never re-examined. Best practice is to re-document the criteria at the annual treatment review and at any change in level of care, even when the diagnosis itself does not change.

CPT codes commonly paired with the borderline-type code

The diagnostic code identifies the condition; the procedure code identifies the service rendered. The CPT codes most commonly paired with this diagnosis in outpatient therapy are 90791 for the diagnostic evaluation at intake (no medical services), 90834 for psychotherapy of approximately 45 minutes (the most common routine code for individual sessions in BPD treatment), and 90837 for approximately 60 minutes (often the right pick when DBT individual sessions include phone coaching review or diary card review on top of the standard agenda). For DBT skills group, 90853 (group psychotherapy, non-multi-family) is the standard pairing. When family members or significant others are seen with the client for relationship work, 90847 (family therapy with patient present) applies. Add-on code 90785 for interactive complexity may apply when communication factors complicate the session.

The CPT 90791 documentation expectations on the diagnostic evaluation that opens a BPD chart mirror the 90791 conventions used across the rest of the F-code library.

The diagnostic and procedure codes travel together on the claim line. The borderline-type code should appear as the primary diagnosis pointer on each psychotherapy CPT line for the duration of the treatment episode. When comorbid mood, anxiety, trauma, or substance-use codes are also active, they appear as additional pointers, with the focus-of-treatment code listed first. Time-based CPT codes require time documentation in the note that supports the code billed; payers routinely deny 90837 when the note does not establish that the session ran to the time threshold, and BPD sessions that run long for clinical reasons need that time documentation to hold the higher-level code.

How Emosapien handles the BPD chart during the session

Emosapien’s Scribe Agent works alongside the clinician as an active co-therapist during intake and ongoing sessions. When the conversation establishes a pervasive pattern across the four BPD domains, surfaces at least five of the nine criteria with concrete client examples, and rules out the major differential diagnoses, the agent surfaces the borderline-type code as the diagnostic candidate alongside the criterion checklist, the differential picture, and the suicide-risk reassessment prompt. The clinician reviews the suggestion, accepts or revises it, and the Assessment section of the progress note populates with the criterion-by-criterion mapping, the differential work shown, the risk reassessment, and the treatment plan tied to the personality-level modality in use.

For ongoing sessions, the agent carries the active suicide-risk frame forward, prompts a brief reassessment at each contact, and flags between-session events (self-injurious behaviour, suicidal gestures, safety plan revisions) for explicit documentation in the next note. At the annual review, the agent surfaces the prompt to re-document the criteria so the chart does not drift into the indefinite-carry pattern that auditors flag. This is not coding automation. The clinician makes the diagnosis. The agent surfaces the candidate, shows its work against the DSM-5 criteria and the ICD-10-CM descriptor, holds the differentials and the risk frame in view, and produces an Assessment that is ready for chart review without retyping. See the AI clinical notes overview for how the Scribe Agent handles documentation across the rest of the note, or start a trial to see candidate diagnostic suggestions in your own intake workflow.