Psychotherapy CPT Codes: 90837 vs 90834 vs 90832 (A Coding Guide)

Time-based coding is one of the quietest revenue leaks in a behavioral health organization. The three core individual-psychotherapy codes (90832, 90834, and 90837) differ only by how long the session ran. Yet the wrong choice triggers a denial, a downcode, or a clawback on audit. Multiply a $30–$50 reimbursement gap across thousands of sessions a year and the "small" coding question becomes a six-figure one.

For a treatment center running individual therapy across dozens of clinicians and a dozen payers, the failure mode is rarely one biller who doesn't know the codes. It's the gap between what the clinician documents and what the time-based code requires. Picture a 53-minute session billed as 90837 with no start/stop time in the note. Or a 90832 that should have been 90834. Or a 90837 that a payer silently pays at the 90834 rate. Every one of those is preventable upstream, before the claim leaves the building.

This guide breaks down the three core codes by their time ranges, the add-on and related codes you'll bill alongside them, and exactly what documentation supports each. It also covers why 90837 gets downcoded to 90834 and how to defend the longer session, plus where AI/agentic systems catch time-vs-code mismatches before submission. Throughout, every threshold and benchmark is sourced inline. And where a rule varies by payer, we say so rather than pretend there's one universal answer.

The Three Core Psychotherapy Codes and Their Time Ranges

Individual outpatient psychotherapy with a patient is coded by time spent face-to-face (or synchronous audio-video for telehealth). CPT defines three timed codes, each tied to a typical duration. Because exact session lengths vary, CPT applies the "midpoint" rule for time-based codes. You may report a timed code once the time spent passes the midpoint between it and the next-lower option. In practice the field uses three working ranges.

Sources: AMA - CPT Evaluation and Management / time-based reporting guidance · BreezyBilling - 90834 vs. 90837

A few things this table makes clear:

• There is no code for a session under 16 minutes of psychotherapy. A contact that brief generally isn't separately reportable as psychotherapy. It may fold into an E/M service or simply not meet the threshold to bill.
• The "typical time" in the descriptor is not the floor. A 90834 is described as "45 minutes," but you may report it from 38 minutes up. And you must report it (not 90837) up through 52 minutes. The descriptor time is a label, not the billing requirement.
• 53 minutes is the gate to 90837. This is the single most consequential number in this guide. One minute short and the correct code is 90834. This is also why undocumented time is so dangerous. A payer can't tell a 55-minute session from a 50-minute one without a start/stop time, and on review it will assume the lower-paying code.

These three codes describe psychotherapy delivered without a medical evaluation and management service on the same date by the same provider. When a psychiatrist or PMHNP delivers both medication management and psychotherapy in one visit, the psychotherapy is reported with an add-on code (90833, 90836, 90838) layered on top of an E/M code. That's a different scenario, covered in the next section.

Add-On and Related Codes You'll Bill Alongside Them

The three core codes rarely live alone. Behavioral health claims routinely pair them with add-on codes for complexity, crisis, or family involvement. Getting those pairings right is where a lot of clean-claims rate is won or lost.

Sources: AMA - CPT code set · BreezyBilling - 90834 vs. 90837

Three practical notes for an RCM team:

90785 (interactive complexity) is for the added work of communication barriers: a young child, an interpreter or caregiver involved in care, mandated reporting that complicates the session, or high emotional reactivity that impedes the therapeutic exchange. It is never a standalone charge. It rides on 90832/90834/90837 (or the add-on psychotherapy codes). The most common 90785 denial is billing it without a valid primary code, or billing it routinely on nearly every claim. That invites a payer audit, because it signals the modifier is being used as a rate-bump rather than a clinical reality.

The add-on psychotherapy codes (90833/90836/90838) exist because you cannot bill a standalone 90834 and an E/M on the same encounter for the same provider. When both happen, the E/M reflects the medical work and the add-on reflects the psychotherapy time. Counting the same minutes toward both is a classic audit finding.

Crisis codes (90839/90840) require genuine crisis: an urgent assessment and management of a patient in high distress or at risk. They reimburse more than routine psychotherapy and therefore draw scrutiny. Reserve them for documented crisis presentations, not simply long or emotionally intense routine sessions.

Documentation That Supports Each Code

A code is only as defensible as the note behind it. For time-based psychotherapy, payers look for a specific set of elements. And the single most common reason a 90837 gets knocked down to 90834 is a note that never recorded how long the session ran.

Here's what supports each tier:

Sources: HFMA - Navigating Medical Necessity Denials · Supahealth aggregate data from 200+ behavioral health practices.

The pattern is straightforward: as the code moves up in time and reimbursement, the documentation bar rises with it. A 90832 needs the basics. A 90837 needs the basics plus an explicit start/stop time plus a reason the session clinically required an hour rather than 45 minutes.

A note that reads "53 minutes (2:00–2:53 PM)" at the top removes the single biggest argument a payer has for downcoding. A note that says only "Individual therapy, continued CBT" gives the payer everything it needs to assume the lower code. In time-based coding, time is not a clerical detail. It is the billable fact.

Why 90837 Gets Downcoded to 90834 - and How to Defend It

90837 is the most-watched routine psychotherapy code in behavioral health. It pays more than 90834, often $20–$50 more per session depending on the payer and region. So payers have a direct financial incentive to scrutinize it, and many run utilization programs specifically around it. Some have historically sent letters to providers who bill 90837 at a high frequency, asking them to "justify" the longer session or shift toward 90834.

There are two distinct ways the money disappears:

1. A formal denial. The claim is rejected for insufficient documentation of medical necessity or time, and you have to appeal.
2. A soft denial / silent downcode. The claim is paid, but at the 90834 rate. No rejection notice, just less money. This is the dangerous one, because nobody notices unless they're comparing expected vs. actual reimbursement by CPT code.

How to defend a legitimate 90837:

• Record the start/stop time on every session. Non-negotiable. "2:00–2:55 PM (55 min)" ends the time argument before it starts.
• Document why an hour was clinically necessary. Think complex trauma processing, a safety assessment that extended the session, a high-acuity presentation, or intensive exposure work. The note should make a 45-minute session feel insufficient for what the patient needed that week.
• Anchor to objective severity. A GAD-7 of 16 or a PHQ-9 of 19 with documented functional impairment supports a longer, more intensive session far better than "patient still anxious."
• Don't bill 90837 by default. If your practice bills 90837 on nearly every session regardless of length, you become an audit target. You also weaken your defense for the sessions that genuinely ran long. Code to the actual clock, every time.
• Know your payer's posture. Some payers and Medicaid MCOs are aggressive on 90837; some are neutral. Track your downcode rate per payer so you know where to tighten documentation first.

The honest framing: 90837 is not a code to avoid. It's a code to earn with the clock and the clinical reason, session by session. Centers that document time and necessity consistently keep their 90837 revenue. Centers that bill it reflexively lose it on audit.

This scrutiny is also part of a larger pattern. Payers are increasingly using automated systems to review and adjudicate claims at scale, including utilization rules that flag high-frequency 90837 billing automatically. A treatment center can't out-argue an algorithm one appeal at a time. The durable answer is to get the documentation right before the claim is submitted, so there's nothing to flag.

Telehealth, POS, and Modifier Implications

The same three codes (90832/90834/90837) are used for telehealth. The modality doesn't change the code, but it changes the place of service (POS) and modifier requirements, and those are a major source of avoidable denials.

Sources: CMS - Telehealth Services / place-of-service guidance · 24/7 Medical Billing Services - Telehealth Billing: POS 02 vs 10 & Modifier 95

The headache here is that telehealth rules are payer-specific and change frequently. Some commercial payers want POS 02 with modifier 95. Some want POS 10 for home-based care. Some Medicaid programs have their own conventions and revise them mid-year. Federal telehealth flexibilities have repeatedly been extended on rolling deadlines, so what's covered, and how it's coded, shifts with policy cycles.

The takeaway for an RCM team: don't assume a single telehealth coding pattern works across your whole payer mix. The combination of correct code (90834/90837) + correct POS (02 vs 10) + correct modifier (95 vs 93) has to be verified per payer, and re-verified when policies turn over. A claims-scrubbing system that holds payer-by-payer telehealth rules is the only way to keep this straight at volume.

Payer-Specific Session Limits and Scrutiny

Beyond the per-claim coding question, two payer-level realities shape psychotherapy billing at a treatment center: session limits and uneven scrutiny.

Session limits. Many plans cap the number of psychotherapy sessions they'll cover in a benefit year, or require prior authorization beyond a threshold. The exact number varies widely. Some plans set soft limits that trigger a medical-necessity review; others require auth after a set count. Behavioral health denial rates run 15–25% nationally, roughly two to three times the 5–10% medical average, and a meaningful slice of that comes from sessions billed past a limit nobody flagged in advance.

Sources: KFF - Claims Denials and Appeals in ACA Marketplace Plans (2024) · ICANotes - Behavioral Health Billing Metrics & KPIs (2025)

The fix is upstream: surface the session limit and prior-auth threshold during benefits verification, then track each patient's session count against it so the front desk and billing team see visit 18 of a 20-session cap coming, not after the denial lands.

Uneven scrutiny. Not every payer watches the same codes the same way. As covered above, 90837 draws extra attention. Mental health parity enforcement is uneven, and behavioral health claims are scrutinized more aggressively than comparable medical claims. Payers are also industrializing this scrutiny: automated adjudication lets a payer review and reject claims at machine speed and scale. ProPublica documented one insurer's system that let its doctors reject more than 300,000 claims over two months, averaging about 1.2 seconds per claim, without opening patient files.

Sources: ProPublica - How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them (2023)

The asymmetry is the point. When payers adjudicate at machine scale and providers still appeal by hand, the math favors the payer. Not because the denials are right (most appealed denials are overturned), but because most denials are never appealed at all. The durable answer isn't to appeal faster. It's to get the code, the time, the documentation, and the session count right before submission, so far fewer claims ever enter the dispute machine.

AI and Agentic Systems for Time-Based Coding

AI in RCM has moved past "better OCR." What's emerging now is agentic systems that handle entire workflows end-to-end, with human review at the edges. For time-based psychotherapy coding specifically, the workflow that matters is narrow and rules-heavy. It's exactly the kind of work that breaks at volume, because no human biller can hold every payer's time table, telehealth convention, and session limit in working memory across thousands of claims a month.

That's the job of the claims agent inside Supabill. It holds state-by-state and payer-by-payer claim guidelines in its core database and scrubs every claim against those rules in real time, before submission. For the codes in this guide, that means catching the mismatches that drive downcodes and denials:

• Time-vs-code mismatches. When a note documents 47 minutes but the claim carries 90837, the agent flags it in real time. The session falls in the 38–52 minute band, so the defensible code is 90834. Conversely, when a 55-minute session is coded 90834, it can surface the under-coding so the center captures the 90837 it earned.
• Missing time documentation. A 90837 with no start/stop time in the note is the single most common reason the code gets downcoded. The agent flags the gap before the claim leaves, so the clinician adds "2:00–2:55 PM" rather than losing the appeal three weeks later.
• Add-on and modifier errors. A 90785 with no valid primary code, a telehealth 90834 missing modifier 95, POS 02 where the payer migrated to 10: each caught against the payer's current rule set.
• Session-limit approaches. When the benefits verification agent surfaces that a payer caps psychotherapy at 20 sessions a year, the claims agent flags that patient's submissions as they approach the limit. The center secures authorization before visit 21, not after the denial.

Honest limits. An agent can't override a payer's policy change. When a payer moves a code to non-covered or revises its telehealth POS rule, the system catches the change and stops submitting against the old rule, but it can't make the payer pay. It can't make a clinically thin 25-minute check-in into a billable 90837, and it won't help if the underlying note never captured the work. What it does fix is the high-volume, rules-based checking (time vs. code, modifier, POS, session count) that humans get wrong simply because there's too much of it to track perfectly across every payer and every state.

If you're interested, book a demo here to learn more.

Quick Wins

Things an RCM lead can put in place this week:

1. Make start/stop time a required note field for every individual-therapy session. Not optional. This single change removes the biggest argument payers have for downcoding 90837 to 90834.
2. Run an expected-vs-actual reimbursement report by CPT. Compare what you billed for 90837 against what you were paid. Silent downcodes show up here and nowhere else.
3. Audit your last 30 90837 claims. Confirm each has a documented time of 53+ minutes and a clinical reason for the hour. Fix the pattern, not just the claims.
4. Map session limits for your top 5 payers and start tracking each patient's session count against them. Catch visit 18 of 20, not the denial at visit 21.
5. Verify your telehealth coding per payer: code + POS (02 vs 10) + modifier (95 vs 93). Don't assume one pattern works across the whole mix.
6. Check your 90785 frequency. If it's on nearly every claim, you're an audit target. Reserve it for genuine interactive complexity.

FAQ

Q: What's the exact time range for 90832, 90834, and 90837?

A: Using the standard midpoint convention for these timed codes: 90832 covers 16–37 minutes, 90834 covers 38–52 minutes, and 90837 covers 53 minutes or more. The "typical" times in the CPT descriptors (30/45/60) are labels, not the billing floor. You bill to the actual time within these bands. Always confirm against your contracted payer, since a minority publish their own time tables.

Q: Can I bill 90837 for a 50-minute session?

A: No. Fifty minutes falls in the 38–52 minute band, so the correct code is 90834. You need to reach 53 minutes of face-to-face psychotherapy time to report 90837. Billing 90837 for a 50-minute session is exactly the pattern that triggers downcodes and clawbacks on audit.

Q: Why does 90837 keep getting paid at the 90834 rate?

A: That's a silent downcode (soft denial). The claim is paid, but at the lower rate, with no formal rejection. It usually happens when the note lacks a documented start/stop time or a clear medical-necessity rationale for the longer session, or because the payer applies a utilization rule to 90837. Track expected vs. actual reimbursement by CPT to catch it, and document time on every session to prevent it.

Q: Is it safe to bill 90837 regularly, or will it trigger an audit?

A: It's safe when it reflects reality. The risk isn't billing 90837. It's billing it reflexively on nearly every session regardless of length. That pattern flags utilization review. Code to the actual clock: bill 90837 when the session genuinely ran 53+ minutes and the note supports it, and bill 90834 when it didn't. Consistent, accurate coding is more defensible than a uniform high-code pattern.

Q: What documentation do I need to defend a 90837?

A: At minimum: a start/stop time (or total minutes) showing 53+ minutes, the presenting problem with severity (ideally a measure like PHQ-9 or GAD-7), functional impairment, the specific interventions used, treatment-plan linkage, and the piece unique to 90837, a reason the longer session was clinically necessary that week. The note should make a 45-minute session feel insufficient for what the patient needed.

Q: Can I bill 90785 (interactive complexity) on its own?

A: No. 90785 is an add-on code. It must accompany a primary psychotherapy code (90832/90834/90837 or the add-on psychotherapy codes). It captures the extra work of communication barriers, such as involving a young child, an interpreter, or a caregiver in care. Billing it without a valid primary code is a guaranteed denial, and billing it on nearly every claim invites an audit.

Q: How is psychotherapy coded when it's done with medication management on the same day?

A: You don't use the standalone 90832/90834/90837 codes. Instead, the medical work is reported with an E/M code (e.g., 99213/99214) and the psychotherapy is reported with an add-on code: 90833 (30 min), 90836 (45 min), or 90838 (60 min), layered on top. The time counted for the add-on is psychotherapy time only, kept separate from the E/M work. Counting the same minutes toward both is a classic audit finding.

Q: Do the same time-based codes apply to telehealth?

A: Yes. The code (90832/90834/90837) is the same for telehealth. What changes is the place of service (POS 02 for non-home, POS 10 for the patient's home) and the modifier (95 for audio-video, 93 for audio-only where covered). These requirements are payer-specific and change frequently, so verify per payer rather than assuming one telehealth pattern works everywhere.

Q: What's the difference between 90839 and 90837?

A: 90837 is routine 60-minute psychotherapy. 90839 is psychotherapy for crisis, an urgent assessment and management of a patient in high distress or at risk, with 90840 as the add-on for each additional 30 minutes. Crisis codes reimburse more and draw scrutiny, so reserve them for documented crisis presentations, not simply long or emotionally intense routine sessions.

Q: When do I use 90846 vs 90847 for family therapy?

A: 90847 is family psychotherapy with the patient present; 90846 is family psychotherapy without the patient present. Both require documentation of why family work is medically necessary for the identified patient's treatment. 90847 is among the most-scrutinized family codes, so make the medical-necessity rationale for including the family explicit in the note.

Q: Are there session limits on psychotherapy, and how do I avoid denials from them?

A: Many plans cap covered psychotherapy sessions per benefit year or require prior authorization beyond a threshold, but the specific number varies widely by payer and plan. The way to avoid the denial is upstream: surface the session limit and auth threshold during benefits verification, then track each patient's session count against it so you secure authorization before the cap, not after the denial.

Q: Which of these codes gets denied most often, and why?

A: 90837 is the most commonly downcoded routine code (to 90834), usually for missing time documentation or thin medical necessity. 90847 is frequently denied when the medical necessity for family inclusion isn't documented. 90785 is denied when billed without a valid primary code or used too routinely. Across all of them, the recurring root cause is the same: the note doesn't carry the specific element (time, necessity, or a valid primary) that the code requires.

References

• American Medical Association. (2026). CPT (Current Procedural Terminology). https://www.ama-assn.org/practice-management/cpt
• American Medical Association. (2025). Prior Authorization Physician Survey. https://www.ama-assn.org/practice-management/prior-authorization
• BreezyBilling. (2025). 90834 vs. 90837. https://www.breezybilling.com/blog/90834-vs-90837
• BreezyBilling. (2025). Timely Filing Denials in Behavioral Health. https://www.breezybilling.net/blog/timely-filing-denial
• CareCloud Continuum. (2025). How Much Are Denials Really Costing You? https://carecloud.com/continuum/how-much-are-denials-really-costing-you/
• Centers for Medicare & Medicaid Services. (2026). Telehealth Services. https://www.cms.gov/medicare/coverage/telehealth
• Centers for Medicare & Medicaid Services. (2024). Prior Authorization and Pre-Claim Review Program Statistics, FY 2024. https://www.cms.gov/files/document/pre-claim-review-program-statistics-document-fy-24.pdf
• Cipher Billing. (2025). Behavioral Health Claims Trends in 2025. https://cipherbilling.com/behavioral-health-claims-trends-in-2025-what-providers-need-to-know/
• Global AHS. (2025). Behavioral Health Billing Denial Rate Benchmarks. https://www.globalahs.com/behavioral-health-billing-denial-rate-benchmarks/
• Healthcare Dive. (2025). More insurance claims denials are being overturned upon appeal. https://www.healthcaredive.com/news/insurance-denials-overturned-appeal-new-york-study-JAMA/817490/
• HFMA. (2025). Navigating Medical Necessity Denials. https://www.hfma.org/revenue-cycle/navigating-medical-necessity-denials-strategies-for-successful-resolution/
• HFMA. (2025). Navigating the Rising Tide of Denials. https://www.hfma.org/revenue-cycle/denials-management/navigating-the-rising-tide-of-denials/
• ICANotes. (2025). Behavioral Health Billing Metrics & KPIs 2025. https://www.icanotes.com/2025/09/24/behavioral-health-billing-metrics-kpis/
• KFF. (2025). Claims Denials and Appeals in ACA Marketplace Plans in 2024. https://www.kff.org/patient-consumer-protections/claims-denials-and-appeals-in-aca-marketplace-plans-in-2024/
• KFF. (2025). Medicare Advantage Insurers Made Nearly 53 Million Prior Authorization Determinations in 2024. https://www.kff.org/medicare/medicare-advantage-insurers-made-nearly-53-million-prior-authorization-determinations-in-2024/
• ProPublica. (2023). How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them. https://www.propublica.org/article/cigna-pxdx-medical-health-insurance-rejection-claims
• SimiTree. (2025). Behavioral Health Clean Claims Rate. https://simitreehc.com/simitree-blog/how-to-calculate-and-improve-your-behavioral-health-clean-claims-rate/
• 24/7 Medical Billing Services. (2026). Telehealth Billing 2026: POS 02 vs 10 & Modifier 95. https://www.247medicalbillingservices.com/blog/telehealth-billing-in-2026-solving-the-pos-02-vs-10-modifier-95-confusion
• Supahealth. (2026). Aggregate billing and denials data from 200+ behavioral health practices. Internal dataset.