July 19, 2026
11 min read
By Albert Wong, PhD · Clinical Psychologist
The short answer
Bill 90834 for 38–52 minutes of face-to-face psychotherapy and 90837 for 53 minutes or more. The clock counts therapy time only — it stops when the client leaves, not when the note is done. The difference is real money: roughly $114 vs. $167 per session under 2026 national Medicare non-facility rates, about $50 every time. 90837 is not an audit trap; it's a standard code that requires the session to actually run 53+ minutes and a note that says so. Write start and stop times in every note and bill the code that matches the clock.
It's 4:50 on a Thursday. The session started at 3:55 and ended at 4:52 — fifty-seven minutes, because that's how long the work took. Now you're at the claim screen with two codes in front of you. One is the honest one. It also happens to be the one that pays about fifty dollars more, and somewhere in the back of your head a colleague's voice is saying I heard people who bill that one get letters.
So you hover. Maybe you click 90834 — again — for a session that wasn't a 90834. You just donated fifty dollars to one of the largest insurance companies in America, out of fear of a letter you've never actually seen. Let's fix that. The rules here are short, the math is blunt, and the fear has a specific history you deserve to know.
The individual psychotherapy codes are timed codes, and the ranges are not folklore — they come from CPT's midpoint convention. The codes are named for 30, 45, and 60 minutes, and you bill the one whose target your session time is closest to. The midpoint between 45 and 60 is 52.5, which rounds to 53 — so 53 minutes is the line where a session stops being a 90834 and becomes a 90837.
| Code | Named for | Actual time range |
|---|---|---|
| 90832 | 30-minute psychotherapy | 16–37 minutes |
| 90834 | 45-minute psychotherapy | 38–52 minutes |
| 90837 | 60-minute psychotherapy | 53 minutes or more |
Three things about that clock, because this is where claims live or die:
Notice what this means in practice: the traditional "50-minute hour" is a 90834. If your sessions truly run 53 to 60 minutes of clinical contact — many therapists' do — they're 90837s. The code follows the clock. That's the whole doctrine.
One more thing before the money: telehealth changes none of this. A video session uses the same codes and the same time thresholds — 53 minutes on camera is a 90837, same as 53 minutes on the couch. What changes is the place-of-service code and, for many payers, a telehealth modifier on the claim line. The clock rules stay the clock rules.
As of 2026, national Medicare non-facility rates — before geographic adjustment, so your locality will differ — run roughly $114 for 90834 and $167 for 90837. Call it a $50-and-change difference per session. Commercial payers generally pay more than Medicare in absolute terms, and the gap between the two codes varies by contract, but a $25–60 spread per session is a realistic planning range.
Now do what the fee schedule hopes you won't, and multiply. Say you see twenty clients a week, your sessions genuinely run 55 minutes, and — out of letter-fear — you bill every one of them as a 90834. Twenty sessions a week, forty-six working weeks: 920 sessions a year. At the Medicare-rate gap of roughly $53, that's around $48,000. Not a rounding error. Not a "nice to have." A salary, left on the payer's side of the table for work you already did. Even at half that caseload, it's a family vacation and the quarterly taxes. Undercoding out of anxiety isn't caution. It's a donation.
(If you want the full journey of what happens to that claim after you pick the code — eligibility, submission, ERA, patient responsibility — we walk the whole pipeline in insurance billing, explained.)
Here's the ghost story, told straight. Years ago, Optum — United's behavioral health arm — began mailing letters to therapists whose billing showed a much higher share of 90837 than their peers. The letters weren't denials. They weren't accusations of fraud. They were outlier notices: we noticed your pattern is unusual; be sure your documentation supports it. But a letter from a payer with the word "review" in it lands in a solo practice like a foghorn in a quiet harbor, and the profession absorbed exactly the wrong lesson: that 90837 itself is dangerous. That it's a "red flag code." That safe therapists bill 90834.
None of that is true. 90837 is a standard, valid, ordinary CPT code, and payers pay it every day without comment. What utilization-review software actually screens for is pattern and support. A clinician billing 90837 for one hundred percent of sessions, every client, every week, with notes that contain no times and read like ten-minute check-ins — that pattern draws a second look, because it suggests the code was picked by template rather than by clock. A clinician whose sessions run long because that's how they practice, whose notes say 4:00–4:56, and whose documentation reflects an hour of actual clinical work? That clinician can bill 90837 all year and sleep fine.
So the plain rule, the one that survives every payer's weather: bill the code that matches the clock, and write the clock down. Not the code that feels safe. Not the code that pays best. The one the minutes say. Your documentation is the ballast — with it, an outlier letter is a squall: loud, brief, survivable. Without it, the same letter is a real problem. One practical footnote: a few plans have historically required prior authorization for 90837 specifically. Ask when you verify benefits, not after the claim bounces.
If a reviewer ever does open your chart, three things decide how that afternoon goes:
Keep those three current and you can bill exactly what happened, every time, with the calm of someone whose log matches their voyage.
Every risk in this article is really one risk: drift between what the note says and what the claim bills. A 90837 claim over a note with no times. A 55-minute session billed as 90834 because the claim screen defaulted there. The fix isn't vigilance — vigilance runs out around the tenth claim of the night. The fix is a system where the note's session time and the claim's code come from the same source. In Practice Harbor, the note carries the session's start and stop times, and the claim inherits the code the clock supports — so the documentation and the billing can't quietly sail apart. You pick honest minutes once; everything downstream agrees with you.
Session times captured in the note, claims that inherit the right code, and a billing pipeline built for solo practices. Free for pre-licensed clinicians, $19/mo licensed.
At least 53 minutes of face-to-face psychotherapy. The threshold comes from CPT’s midpoint convention: 90837 is named for 60 minutes, 90834 for 45, and the midpoint between them (52.5) rounds to 53. Only direct therapy time counts — documentation, scheduling, and copay collection are excluded — so a 52-minute session is a 90834 and a 53-minute session is a 90837.
No — 90837 is a standard, valid code that payers reimburse every day. The fear traces to outlier letters (most famously from Optum) sent to clinicians whose share of 90837 was far above peers; those were informational notices, not denials. What draws payer review is a pattern that documentation cannot support: billing 90837 universally while notes lack session times or content consistent with 53+ minutes. Bill the code that matches the clock and document the time, and 90837 is as safe as any other code.
Under 2026 national Medicare non-facility rates, roughly $114 for 90834 versus about $167 for 90837 — around a $50 difference per session before geographic adjustment. Commercial payers typically pay more than Medicare, with the gap between the codes varying by contract; $25–60 per session is a realistic planning range. Over a full caseload, routinely undercoding 53+ minute sessions as 90834 can cost tens of thousands of dollars a year.
Yes — record start and stop times (or total face-to-face time) in every psychotherapy note, for every timed code, not just 90837. The individual psychotherapy codes 90832, 90834, and 90837 are selected by session length, so the time in the note is what substantiates the code on the claim. Notes without times are the most common documentation gap payers cite when reviewing timed psychotherapy claims.