Borderline personality disorder ICD 10 code F60.3

TL;DR: The borderline personality disorder ICD 10 code is F60.3, and it is billable. It also carries three legacy synonyms, including emotionally unstable personality disorder, which is what you will see on records from other systems. This is the diagnosis where documentation quality has consequences beyond billing.

Key takeaways

• F60.3 is billable and covers borderline personality disorder.
• The code lists emotionally unstable, aggressive, and explosive personality disorder as descriptors.
• NIMH data does not support the common belief that this diagnosis is sex-linked.
• Criteria-level documentation matters more here than in almost any other diagnosis.
• Your notes will be read by other clinicians, and how you write them shapes the care your client receives next.

There is a diagnosis clinicians hesitate over more than any other. Not because the criteria are unclear, but because of what happens to a client after it lands on their record.

That hesitation is understandable and it produces a specific documentation failure. Charts describing all nine criteria in detail while coding something vaguer, or charts carrying F60.3 with almost nothing behind it because the clinician did not want to write it down. Both create problems, and the second creates them for the client.

What is the ICD-10 code for borderline personality disorder?

The code is F60.3, with the official ICD-10-CM descriptor "Borderline personality disorder," and it is billable rather than a parent category (AAPC).

The listed additional descriptors are the surprising part. F60.3 also covers "Emotionally unstable personality disorder," "Aggressive personality (disorder)," and "Explosive personality (disorder)" (AAPC).

That matters practically. Emotionally unstable personality disorder is the standard term in several health systems outside the US, so a client arriving with records from elsewhere may carry that label for what your system calls borderline personality disorder. They are the same code. You do not need to re-diagnose to reconcile the terminology, though you do need to satisfy yourself the diagnosis fits.

The aggressive and explosive descriptors are historical residue and rarely used clinically. Knowing they sit under F60.3 saves confusion when an old record surfaces them.

Does the prevalence data match what clinicians assume?

Not on sex, which is the assumption most worth examining. An estimated 1.4% of US adults had borderline personality disorder in the past year, within an overall personality disorder prevalence of 9.1% (NIMH).

Here is the finding that should give any clinician pause. NIMH reports that "sex and race were not found to be associated with the prevalence of personality disorders" (NIMH).

The widespread clinical impression that borderline personality disorder is predominantly a female diagnosis is not supported by that population data. What the impression more likely reflects is differential help-seeking, differential referral, and differential diagnosis, meaning men presenting with similar difficulties may be receiving different labels.

The practical implication is uncomfortable and worth sitting with. If your own caseload shows a heavy sex skew on F60.3, the population data suggests looking at your diagnostic process rather than treating the skew as expected. That is not a comfortable audit to run. It is a useful one.

Why does documentation quality matter more for this diagnosis?

Because the record travels, and it changes how the next clinician behaves before they have met your client. A personality disorder diagnosis attracts assumptions in a way an anxiety diagnosis does not, and a thin chart leaves those assumptions unchallenged.

Three practices make a real difference.

Document criteria, not impressions. "Client is very borderline" is not a clinical statement, and it is not defensible. Which criteria, evidenced how, over what period.

Describe behavior rather than character. "Client became angry and left the session after I raised the ending date, returning the following week to apologize" is an observation. "Client is manipulative" is a judgment that tells a reader about the writer.

Record strengths and function. A chart containing only the pathology gives the next clinician a distorted picture and a worse starting point. Note what the client manages well, since it is clinically true and it is also the part most likely to be omitted.

This is the diagnosis where the gap between defensible documentation and good documentation is widest. A note can satisfy a payer while doing real harm to the person it describes.

For structuring the broader formulation around a diagnosis like this, our case conceptualization guide and the 5 Ps framework both give you something more useful than a criteria checklist.

What does a defensible F60.3 note look like?

It ties observed behavior to named criteria, spans time rather than a single session, and separates what you saw from what you inferred.

Weak: Client was dysregulated again today. Typical presentation. Discussed boundaries.

Defensible: Client arrived 20 minutes late and initially declined to engage, stating the clinician "clearly doesn't care." This followed the previous session's discussion of a two-week break for annual leave. Within the session, she moved from anger to distress to warmth over approximately 25 minutes. She independently named the pattern for the first time in treatment, linking it to fear of abandonment, which she has previously been unable to articulate. No self-harm reported since the last session, extending the current interval to six weeks. Interventions: validation, chain analysis of the arrival, and review of the crisis plan.

The second version records affective instability and abandonment sensitivity as behavior with a trigger and a sequence, notes a genuine therapeutic gain, and includes the risk interval. A reader learns what happened and what is improving, which is what a clinical record is for.

Note the risk line especially. With this diagnosis, an absent risk statement is read as an omission rather than as an absence of risk.

When would you use F60.3, and when would you not?

A personality disorder diagnosis requires a pervasive, enduring pattern across contexts and time, not a presentation during a crisis or an episode. This is the diagnosis where premature coding does the most harm, because the label persists in records long after the clinical picture has moved on.

Use F60.3 here

A client in her late twenties has a documented pattern going back to adolescence: intense unstable relationships, recurrent self-harm during periods of perceived rejection, chronic emptiness, rapid shifts between idealizing and devaluing people including previous therapists, and impulsive spending. The pattern is present at work, in friendships, and in treatment, and it has been observed across eighteen months of your own notes.

Code F60.3. Multiple criteria, evidenced across contexts, stable over years rather than weeks, and observed directly rather than inferred from a single account.

Do not use F60.3 here

A client presents in acute crisis after a relationship ended. She is dysregulated, has self-harmed for the first time in years, and is expressing intense fear of being alone.

Do not code a personality disorder from a crisis presentation. What you are seeing may be the pattern or may be an acute response to a significant loss. Code what you can support now, and reassess when the crisis has settled.

A client in a severe depressive episode presents with emotional instability, emptiness, and irritability. These features were not described before the episode began.

Do not code personality on state. Several criteria overlap with an active mood episode, and attributing them to personality while the episode is untreated is a well-recognized error. Treat the depression and reassess once the mood has stabilized.

A sixteen-year-old has six months of emotional instability, conflict with parents, and one episode of self-harm.

Exercise real caution. Personality is still developing, the duration is short, and the consequences of this label following a young person through the health system are significant. Document the observed difficulties without assigning a personality disorder code on this evidence.

A client presents with emotional dysregulation, relationship difficulty, dissociation, and a persistent negative self-concept, all following prolonged childhood abuse.

Consider whether a trauma diagnosis accounts for the picture more completely. The presentations overlap substantially, and defaulting to a personality diagnosis where a trauma formulation fits better shapes both the treatment offered and the way every future clinician reads the chart.

How the criteria-level evidence can be automated

Personality disorder documentation is a longitudinal problem. No single session evidences the diagnosis, because the criteria are patterns across time and context. That makes it a poor fit for memory and a reasonable fit for a system that carries detail forward.

The mechanism, concretely.

Behavior lands as behavior. The failure mode here is characterological shorthand, and it happens under time pressure at the end of a day. A note drafted from the session captures the sequence, the trigger, and the client's own words, which is precisely what stops "manipulative" appearing in a record where "asked three times whether I was planning to end therapy" belongs.

The pattern becomes visible across sessions. Criteria are established by recurrence. When each session's detail is retained rather than compressed, the recurrence is legible in the chart instead of living in the clinician's head.

The risk interval stays continuous. Six weeks without self-harm is a meaningful clinical fact and it only exists if every intervening note recorded the status. One missing entry breaks the count.

Supanote drafts each session note from what happened, so the behavioral specifics and the risk status stay in the record rather than being smoothed into summary. You review, edit, and sign every draft, and nothing is filed without you.

The boundary here is sharper than for any other diagnosis in this series. Software must not suggest a personality disorder diagnosis, must not infer criteria from language, and must not characterise a client. Those judgments carry lifelong consequences for the person and they belong entirely to a trained clinician who knows them. Any product marketing automated personality-disorder detection is one to refuse.

What automation legitimately offers is narrower and still valuable: a record that describes what happened rather than what you concluded, which is the kind of record that serves your client when someone else reads it in three years.

Want your notes to describe behavior rather than character? Try for Free.

FAQ

Q: What is the ICD-10 code for borderline personality disorder?

A: F60.3, officially "Borderline personality disorder." It is a billable code and it also carries emotionally unstable, aggressive, and explosive personality disorder as listed descriptors, which is why records from other systems may use different terminology for the same code.

Q: Is emotionally unstable personality disorder the same as BPD?

A: For coding purposes, yes. Emotionally unstable personality disorder is listed as a descriptor under F60.3, and it is the standard term in several health systems outside the US. You would still satisfy yourself the diagnosis fits rather than accepting a transferred label uncritically.

Q: Is BPD more common in women?

A: NIMH reports that sex was not found to be associated with the prevalence of personality disorders. The clinical impression of a female-skewed diagnosis more likely reflects differences in help-seeking, referral, and diagnostic practice than differences in underlying prevalence.

Q: Should I tell a client about a personality disorder diagnosis?

A: That is a clinical and ethical judgment rather than a coding one, and practice varies. What is not optional is that the diagnosis on the claim matches the diagnosis in the record, since a discrepancy between what you bill and what you document creates a separate problem.

Q: How do I document without stigmatising language?

A: Write behavior and sequence rather than character. Record what happened, what preceded it, and what the client said, and leave out adjectives describing the person. The test is whether the client could read the note and recognise the events described.

Q: Can I code BPD alongside depression or PTSD?

A: Yes, where both are independently present and separately documented. Comorbidity is common with this diagnosis, and each code needs its own supporting evidence rather than one being inferred from the presence of the other.