90832 vs 90834 vs 90837: choosing and documenting the psychotherapy time codes

By Selam Koch, PMHNP-BC, founder of OneStep Scribe and practicing psychiatric nurse practitioner. Published August 2026.

The three standalone psychotherapy codes differ on exactly one axis: time. 90832, 90834, and 90837 describe the same service, individual psychotherapy, at three durations. That should make them the easiest codes in behavioral health to get right, and in practice they are among the most commonly miscoded, because three details get missed: the real CPT time ranges are wider than the labels suggest, the time has to be the actual face-to-face time rather than the scheduled slot, and the time has to be written in the note.

This guide covers the thresholds, the two accepted ways to document time, how the standalone codes relate to the add-on codes used with an E/M service, why 90837 gets more payer scrutiny than its siblings, and the recurring errors that turn a legitimate session into a denial or an audit finding.

The three codes and their real time ranges

Each code carries a labeled duration, but CPT time conventions assign each one a range. The code you report is the one whose range contains your actual face-to-face psychotherapy time:

Three consequences follow from the ranges:

• Under 16 minutes, none of these codes applies. A 12-minute supportive conversation is real care, but it does not meet the minimum for the lowest of these time-based codes.
• A 40-minute session is 90834, not 90832. Coding by the label alone undercodes sessions that crossed the 38-minute line, and overcodes 50-minute sessions that belong at 90834.
• 90837 starts at 53 minutes, not 60. A session with 55 documented minutes of psychotherapy supports 90837 under CPT conventions. Whether a specific payer applies the CPT ranges exactly this way is worth confirming, because payer policies vary.

What counts toward the total is face-to-face psychotherapy time with the patient, and under current CPT the patient must be present for all or a majority of the service. Family members or other informants can participate within that session, but a session conducted primarily or entirely with family and without the patient is reported with the family psychotherapy codes, 90846 without the patient present or 90847 with the patient present, not with these individual time codes. What does not count: writing the note, reviewing the chart before the session, scheduling, phone tag, and time the patient spent in the waiting room. The clock measures the therapy, not the appointment.

Time must be documented, and there are two accepted ways

This is the part reviewers check first, because it is the easiest thing to check. A time-based code with no time in the note is unsupported on its face; in a records request, the reviewer does not have to argue about clinical content, they simply observe that the element that selects the code is absent. Two formats are widely accepted:

• Start and stop times. "Psychotherapy 2:04 pm to 2:52 pm." This is the stronger format because it is verifiable against the schedule and hard to produce by template. Some payers and some state Medicaid programs explicitly require start and stop times, so if you work with those payers this is not optional.
• Total face-to-face time. "Total psychotherapy time: 48 minutes, face-to-face." CPT generally supports documenting total time, and many payers accept it. It should state the number of minutes and make clear the time was face-to-face psychotherapy, not the appointment length.

Pick one format and use it every visit. The safest habit is to record start and stop times, because they meet the stricter requirement wherever it exists and they anchor the note to the actual encounter.

Standalone codes versus the add-on codes billed with an E/M

90832, 90834, and 90837 are standalone codes: they describe a visit whose billable service is psychotherapy and nothing else. When the same clinician provides an E/M service and psychotherapy in the same encounter, which is the daily reality for most psychiatric prescribers, the psychotherapy is reported instead with the add-on codes: 90833 (16 to 37 minutes), 90836 (38 to 52 minutes), or 90838 (53 minutes or more), attached to the E/M code.

The practical rule of thumb: therapists live in the standalone codes, and prescribers live mostly in the add-ons, dropping into the standalone codes only for visits where no E/M service was performed. The two families never mix in one encounter; you do not bill 90834 alongside 99214, you bill 99214 with 90836.

The add-on world has its own documentation discipline, and it is stricter, because the psychotherapy time must be carved out from the E/M work and supported on its own. A complete annotated example of what that looks like is in the 90833 note example, and the billing rules for the combination, including the 16-minute floor and the common denial patterns, are covered in billing 99214 and 90833 together.

One boundary case worth naming: the initial evaluation. A first visit is generally reported with a psychiatric diagnostic evaluation code, or, for prescribers, often a new-patient E/M, rather than a standalone psychotherapy code; the distinction between the two evaluation codes, and when each applies, is covered in 90791 vs 90792.

Why 90837 draws payer attention

90837 is the highest-paying code of the three, and that alone explains most of the scrutiny. Several commercial payers have, at various times, profiled 90837 utilization, sent letters to clinicians whose 90837 rates sat far above their peers, requested records for routine 90837 claims, or applied prior-authorization or review policies to the code. None of that means 90837 is disfavored or improper. It is a legitimate, frequently appropriate code, and many evidence-based therapies are genuinely built for a full hour. It does mean the code is billed into a headwind, and the documentation should be ready for the records request before it arrives.

Documentation that tends to support a 90837 claim, with the standing caveat that payer policies vary and nothing here guarantees payment:

• An actual documented time of 53 minutes or more, ideally with start and stop times. The single most common 90837 problem in review is a note with no time, or a time under 53 minutes, under the 60-minute code.
• Session content that accounts for the hour. The interventions, themes, and patient responses described should plausibly fill 53-plus minutes. Three sentences of content under a 60-minute claim create a mismatch a reviewer will notice.
• A reason the extended session serves this patient's treatment, visible in the treatment plan or the note: a structured protocol that runs long by design, such as trauma-focused or exposure-based work, follow-up after a recent crisis (a session that is itself primarily crisis psychotherapy is reported with 90839 and 90840 instead), or complexity that shorter sessions have not been able to contain. This is not a required element in CPT, but when 90837 is your routine code, a documented clinical rationale is what separates a defensible pattern from an unexplained one.
• Variation that matches reality. If some sessions genuinely run 45 minutes and get billed 90834, the chart shows a clinician coding by the clock. If every session in the chart is 90837 with identical times, the pattern itself becomes the finding, even when each individual claim was honest.

If a payer has published a policy on 90837, follow it; if you are unsure whether one exists, verify with the payer or, for Medicare questions, your MAC. Coding should follow what actually happened in the room, in both directions: a true 58-minute session should not be downcoded to 90834 out of fear any more than a 45-minute session should be rounded up.

Common errors, and how each one reads to a reviewer

Billing the scheduled time instead of the actual time

The most common error. The calendar says 60 minutes, so the claim says 90837, but the patient arrived 10 minutes late and the session closed 5 minutes early to schedule the next visit. The face-to-face psychotherapy time was about 45 minutes, and the supported code is 90834. Booking length is not service length, and a chart where billed time always equals slot length is one of the patterns utilization review is built to catch.

Double-counting E/M time toward psychotherapy time

For prescribers using the add-on codes, the psychotherapy minutes must be distinct from the E/M work. Time spent reviewing symptoms, adjusting a medication, and discussing side effects is E/M time; it cannot also be counted toward the 16-minute psychotherapy floor. A 45-minute visit documented as 45 minutes of E/M plus 30 minutes of psychotherapy is arithmetic that falsifies itself. The carve-out mechanics are covered in detail in the 99214 with 90833 guide.

Counting non-face-to-face time

Documentation time, chart review, calls to a pharmacy, and coordination after the patient leaves are real work, but they are not psychotherapy time under these codes. Only face-to-face therapy minutes count toward the range.

No time in the note at all

A clinically excellent note with no time element does not support any of these codes, because time is what selects among them. This is the cheapest error to fix: add a session-time line to your note structure and never sign without it.

Time in the note that contradicts the code

A note stating 44 minutes under a 90837 claim is worse than a note with no time, because it documents the mismatch. Before signing, check that the minutes and the code agree; it takes ten seconds.

Frequently asked questions

Is 53 minutes really enough for 90837?

Under the CPT time ranges, yes: 90837 covers 53 minutes and beyond, and a documented 53-minute session falls in its range. Individual payers can and sometimes do apply their own expectations, so if 90837 is a significant part of your billing, confirm how your major payers handle the threshold rather than assuming the CPT convention controls.

Do I need start and stop times, or is total time enough?

CPT is generally satisfied by a documented total time, and many payers accept it. Some payers and Medicaid programs require start and stop times. Since start and stop times meet the stricter requirement wherever it exists and cost you nothing, they are the better default habit.

I am a prescriber. When would I ever use the standalone codes?

Only when the visit contained no E/M service: a pure therapy session with no medication management, assessment, or medical decision making billed that day. In most prescriber workflows that is uncommon, and the psychotherapy you provide alongside medication management belongs on the add-on codes with your E/M. When in doubt, the question is whether you performed and documented a billable E/M service, not how much of the visit felt like therapy.

What are 90836 and 90838?

They are the add-on counterparts to 90834 and 90837: 90836 covers 38 to 52 minutes and 90838 covers 53 minutes or more of psychotherapy performed with an E/M service in the same encounter. The same time ranges, the same documentation rules, plus the requirement that the psychotherapy time be separate from the E/M work.

Can an AI scribe get the time documentation right?

It can capture the raw material, and that is most of the battle. OneStep Scribe listens to the session and drafts the note with the psychotherapy content preserved, the interventions named, and the session structure visible, so the note's content can support the minutes claimed. The clinician confirms the session time and the code; nothing enters the chart until it is reviewed, edited, and signed. What the software will not do is invent minutes, and neither should the note.

This article is educational and reflects one clinician's documentation approach. It is not clinical, legal, or billing advice for any specific patient or claim, and it does not guarantee any outcome or reimbursement. The sessions and times described are fictional. Coding rules, time thresholds, documentation requirements, and review practices vary by payer and change over time; verify any code, threshold, or requirement with the specific payer or your MAC and a certified coder before billing. Clinical judgment stays with the treating clinician. CPT is a registered trademark of the American Medical Association.