Free for a week, then $19 for your first month
Research

We timed 1,116,796 clinical sessions. The 50-minute therapy hour is real

Median session length for 48 clinical specialties from 1,116,796 live recorded encounters. Twelve talk-therapy professions converge on 51 to 57 minutes, psychiatry runs 17.4, family medicine 12.1. Dataset included.

A clock with a coral sweep marking about 54 minutes beside five ascending bars — the range of clinical session lengths from 12 to 54 minutes.

Median session length for 48 clinical specialties, from live recordings captured by Twofold's ambient scribe across 6,989 clinicians over six months. Twelve talk‑therapy professions converge on 51 to 57 minutes. Psychiatry runs 17.4, family medicine 12.1.

1,116,796
Sessions timed
6,989
Clinicians
48
Reportable specialties
53 min
Where the billing band changes
Six months ending July 2026 · live conversation captures only.

Findings

  • Across 652,932 recorded sessions, all twelve talk-therapy professions land between 51.2 and 57.4 minutes median. The 50-minute hour holds. The four largest professions, 572,532 sessions between them, sit inside a single minute: 53.8 to 54.7.
  • Ten of twelve therapy professions have a median session within two minutes of the 53-minute threshold where CPT 90837 begins paying 15 to 25 percent more. At the other boundary, five professions' 25th percentiles land between 38.6 and 39.4 minutes, on the 38-minute floor of the lower code, 90834.
  • The median psychiatry visit is 17.4 minutes across 226,000 sessions: one third of a therapy session, inside the same field of mental health. Its middle 50 percent spans 8.9 to 30.4 minutes, the widest relative spread of any large specialty.
  • The median family-medicine visit contains 12.1 minutes of recorded conversation; urgent care, 5.1. Internal medicine runs 13.2 and pediatrics 10.2. Time-use surveys report 18 to 20 minutes face-to-face; the difference is the silent exam.
  • Lactation consultants log the longest visits of all 48 specialties: 65 minutes median, above every form of psychotherapy measured. 2,508 sessions from 34 clinicians; the longest talk therapy, trauma therapy, tops out at 57.4.

The 50-minute therapy hour, measured across 652,932 sessions

The 50‑minute hour began as a psychoanalytic convention: a full hour on the books, ten minutes reserved for the note. Nobody could say whether clinicians keep it, because nobody was timing the sessions. An ambient scribe times them incidentally: it listens during the visit to draft the clinical note, and the session's length falls out as a by‑product.

We split every talk‑therapy profession and measured each separately. Different training, licensure, client populations. The medians:

Profession

Median (min)

vs 53

Middle 50% (min)

Sessions

Clinicians

Trauma Therapy

57.4

+4.4

51.7 to 62.2

1,581

13

Family Therapy

55.8

+2.8

47.1 to 61.9

2,317

18

Marriage and Family Therapy

54.7

+1.7

47.9 to 60.0

41,461

192

Mental Health Counseling

54.6

+1.6

47.0 to 59.9

232,219

1,491

Psychotherapy

53.9

+0.9

45.1 to 59.6

186,601

1,349

Clinical Social Work

53.8

+0.8

43.6 to 59.5

112,251

703

Social Work

53.3

+0.3

39.3 to 60.5

18,241

106

Clinical Psychology

52.7

-0.3

38.8 to 58.7

27,098

104

Behavioral Health Counseling

51.9

-1.1

41.2 to 58.5

16,007

85

Behavioral Therapy

51.7

-1.3

39.4 to 59.6

3,217

47

Psychology

51.6

-1.4

38.7 to 58.3

1,000

19

Cognitive Behavioral Therapy

51.2

-1.8

38.6 to 57.3

10,939

46

The four largest professions, 572,532 sessions, sit between 53.8 and 54.7. For most rows the middle 50 percent falls inside 44 to 60 minutes. The hour bends; it rarely breaks.

One honesty check
Three of the twelve labels are modalities rather than professions: CBT, behavioral therapy, trauma therapy. Psychology and clinical psychology are one licensure typed two ways, and specialty is a self-reported profile field. So this is one population of talk therapists sliced twelve ways, not twelve independent replications. The convergence is real all the same, and it behaves like a real signal: the two rows that drift highest, family therapy at 55.8 and trauma therapy at 57.4, are also the two smallest samples, which is what noise does.

The 53-minute line: therapy medians cluster at a billing threshold

Something sits exactly at 53 minutes, and it is not a psychoanalytic convention. It is a price change.

Outpatient psychotherapy is billed in time bands with hard edges:

Code

Face-to-face time

Reimbursement

90832

16 to 37 min

base

90834

38 to 52 min

mid

90837

53 min and up

15 to 25% above 90834 — roughly $30 to $35 more on Medicare

The 53‑minute figure is not clinical. It is the AMA's midpoint between the 45‑minute target of 90834 and the 60‑minute target of 90837 — 52.5, rounded up. A 52‑minute session is 90834. A 53‑minute session is 90837. No rounding, and payers audit the boundary.

Now place the twelve medians against those bands:

Range chart of twelve talk-therapy professions sorted by median against the CPT psychotherapy billing bands. Ten of twelve medians fall between 51.2 and 54.7 minutes, straddling the 53-minute boundary between 90834 and 90837. Red ticks mark five 25th percentiles between 38.6 and 39.4 minutes, sitting on the 38-minute floor of 90834.

Two boundaries, two clusters. Medians pile up at the 53‑minute threshold where payment steps up; five 25th percentiles pile up at the 38‑minute floor below which the session falls to the lower code (clinical psychology 38.8, CBT 38.6, psychology 38.7, social work 39.3, behavioral therapy 39.4). Both are visible in quartiles alone; neither needs a model.

Two explanations, one test
Story one: therapy takes about an hour because that is how long the work takes, and the codes describe practice. Story two: sessions land just above 53 because that is where the higher payment starts, and the codes shape practice. Quartiles cannot separate these. The minute-level distribution can, and the prediction is specific: if billing drives duration, the histogram steps at 53 and at 38, with a deficit at 51 and 52. If practice drives it, the curve is smooth through both boundaries. We are running that histogram and will publish it either way, including the version where this section becomes a coincidence.

There is already one argument against the billing story inside this dataset. Physical, occupational and speech therapy bill under the CMS 8‑minute rule, whose unit steps fall at 23, 38 and 53 minutes. If thresholds pulled duration upward everywhere, physical therapy would sit just above 53. It sits at 50.3, mid‑band. Whatever happens at 53 in psychotherapy is not a universal law of threshold‑seeking.

Psychiatry runs on two clocks: short med checks, long evaluations

If the therapy hour is a contract, psychiatry never signed it. It sits inside mental health, and its median is 17.4 minutes across 226,000 sessions from 1,001 clinicians. Most outpatient psychiatry is medication management, and medication management is a decision, not a conversation.

Range chart showing the middle 50 percent of sessions: four therapy professions sit inside bands from roughly 43 to 61 minutes, while psychiatry spans 8.9 to 30.4 minutes.

The mean is 22.7 against a median of 17.4: many short visits plus a long tail of intakes and evaluations. Two visit types sharing one label. Strictly, mean above median with a wide IQR is also consistent with plain right skew; the histogram will confirm the two clocks or kill them.

The 12-minute doctor visit, and why surveys say 18 to 20

Nobody is surprised that the doctor's visit runs short; patients have complained about it for decades. The complaint now has a number. The median recorded family‑medicine visit is 12.1 minutes, internal medicine 13.2, pediatrics 10.2, urgent care 5.1. Time‑use surveys put primary‑care face‑to‑face time nearer 18 to 20 minutes, and the gap is not an error on either side. It is the definition.

What the scribe measures
An ambient scribe captures conversation. A primary-care visit is not one continuous conversation: talk, silent exam, step out, return. Silence and time out of the room are not in the recording. A telehealth therapy session is one unbroken block, captured nearly whole. Read 12.1 as 12 minutes of recorded conversation, not 12 minutes of appointment; the difference from the survey figure is roughly the size of the silent exam.

Every medical specialty here shows a mean well above its median (family medicine 16.1 against 12.1): most visits are quick, a minority run long. The 15‑minute slot survives because typical visits finish under it and complicated ones borrow the difference.

Conversation-bound versus decision-bound: the split that predicts session length

Rank all 48 specialties on one axis and the mental‑versus‑medical framing falls apart. Physical therapy runs 50.3, four minutes off psychotherapy. Occupational therapy 44.8. The longest sessions belong to lactation consultants at 65.

Ranked bar chart of all 48 reportable specialties by median session length, colored by behavioral or medical category. Lactation consulting leads at 65 minutes, followed by the talk therapies at 51 to 57 and physical therapy at 50.3. Psychiatry sits at 17.4, family medicine 12.1, urgent care 5.1 and geriatric medicine 1.9. Both colors appear at both ends of the axis.
Conversation-bound · 45 to 65 minutes
The talking or hands-on work is the treatment
The session ends when the scheduled block ends
Talk therapy, physical and occupational therapy
Lactation consulting, nutrition therapy
Decision-bound · 5 to 20 minutes
The session exists to reach a decision
Examine, assess, prescribe, adjust — then it ends
Family medicine, psychiatry, urgent care
Dermatology, cardiology, orthopedic surgery

Lactation consulting is filed as Medical and is the longest thing we can see; psychiatry is filed as Behavioral and sits nearer urgent care than therapy. The labels are profile fields, not a taxonomy, and they are the wrong axis.

Failure modes: five ways this data can mislead

Every claim above depends on one instrument. These are the five ways it can mislead, in order of how much they worry us.

01
Recording length is not encounter length
The largest threat, and it cuts against the medical specialties. Stop-and-restart behavior, scribes off during exams, and the telehealth split all bias medical visits short relative to therapy: therapy is heavily remote and captured end to end, primary care is mostly in-person and captured in pieces. Until we publish the split, read the cross-specialty comparison as recorded conversation.
02
The short end of the table is measuring recording behavior
Geriatric medicine: 1.9 minutes median across 13,722 sessions. Not office visits; almost certainly bedside check-ins during facility rounds. Case managers 3.9, case coordinators 2.3, chiropractic 5.2 look similar. We do not get to call these rows artifacts and keep family medicine as fact; same instrument.
03
A few high-volume clinicians can move a specialty's median
Geriatric medicine averages 508 sessions per clinician, podiatry 447, family medicine 140. A few high-volume practices can set a specialty's median, so effective sample size is nearer the clinician count than the session count, and adjacent medians like 54.6, 53.9 and 53.8 should be read as ties. Sessions per clinician is a column in the full table.
04
No confidence intervals on any figure
Every number is a point estimate to one decimal with no interval attached. The decimals are precise, not necessarily accurate.
05
One platform's users, not a census of healthcare
This is Twofold's user base: US outpatient, small-practice heavy, weighted toward behavioral health. The largest measurement of recorded session length we know of; not a census.

Not done yet, in order: telehealth versus in‑person split; exclusion counts at the 1‑minute and 4‑hour bounds; clinician‑level medians and between‑clinician variance; confidence intervals; and the minute‑level histogram, which is running now and gets published either way.

Method: how 1,116,796 sessions were measured

Duration here is the actual length of the recorded encounter, captured as a by‑product of drafting the note. Not a self‑report, not a billing increment, not a documentation timestamp.

Parameter

Value

Window

Six months ending July 2026

Sessions

1,116,796 live conversation captures

Clinicians

6,989

Included

Completed, non-deleted live captures only

Excluded

Dictation and typed notes (no meaningful duration)

Bounds

1 minute to 4 hours (drops accidental taps, runaway recordings)

Estimator

Median and IQR via percentile_cont, linear interpolation

Suppression

Under 500 sessions or under 10 clinicians

Specialty

Clinician self-reported profile field

Privacy

Aggregate start/end times only; audio discarded after drafting; no training on patient data

The full dataset: all 48 specialties

Every specialty that passed the thresholds, ranked by volume, including the rows that complicate the story.

Specialty

Category

Sessions

Clinicians

Sessions / clinician

Median (min)

Middle 50% (min)

Mental Health Counseling

Behavioral

232,219

1,491

156

54.6

47.0 to 59.9

Psychiatry

Behavioral

226,000

1,001

226

17.4

8.9 to 30.4

Psychotherapy

Behavioral

186,601

1,349

138

53.9

45.1 to 59.6

Clinical Social Work

Behavioral

112,251

703

160

53.8

43.6 to 59.5

Family Medicine

Medical

46,501

332

140

12.1

6.1 to 21.3

Marriage and Family Therapy

Behavioral

41,461

192

216

54.7

47.9 to 60.0

Clinical Psychology

Behavioral

27,098

104

261

52.7

38.8 to 58.7

Physical Therapy

Medical

23,637

183

129

50.3

32.8 to 58.6

Social Work

Behavioral

18,241

106

172

53.3

39.3 to 60.5

Behavioral Health Counseling

Behavioral

16,007

85

188

51.9

41.2 to 58.5

Geriatric Medicine

Medical

13,722

27

508

1.9

1.3 to 3.6

Internal Medicine

Medical

12,624

109

116

13.2

7.0 to 21.8

Podiatry

Medical

11,185

25

447

12.8

6.0 to 23.0

Cognitive Behavioral Therapy

Behavioral

10,939

46

238

51.2

38.6 to 57.3

Chiropractic

Medical

9,901

39

254

5.2

2.4 to 11.3

Pain Management

Medical

5,461

23

237

8.5

5.6 to 13.1

Speech Therapy

Medical

5,386

70

77

29.0

22.1 to 38.6

Dietitian

Medical

5,130

58

88

29.2

8.9 to 52.0

Substance Abuse Counseling

Behavioral

4,231

52

81

29.3

14.0 to 46.9

Functional Medicine

Medical

4,018

31

130

29.6

18.5 to 46.7

Pediatrics

Medical

4,005

48

83

10.2

5.9 to 16.6

Gastroenterology

Medical

3,709

13

285

13.0

8.4 to 19.0

Nutrition Therapy

Medical

3,586

30

120

35.0

22.5 to 50.5

Behavioral Therapy

Behavioral

3,217

47

68

51.7

39.4 to 59.6

Occupational Therapy

Medical

2,969

70

42

44.8

24.6 to 59.6

Case Managers

Medical

2,895

68

43

3.9

1.6 to 23.1

Dentistry

Medical

2,509

27

93

21.6

9.8 to 36.7

Lactation Consultant

Medical

2,508

34

74

65.0

46.0 to 88.0

Neurology

Medical

2,490

15

166

13.4

7.6 to 20.9

Addiction Medicine

Behavioral

2,486

27

92

21.0

9.2 to 40.2

Urgent Care

Medical

2,413

11

219

5.1

3.3 to 7.8

Family Therapy

Behavioral

2,317

18

129

55.8

47.1 to 61.9

Addiction Psychiatry

Behavioral

2,117

13

163

19.6

11.1 to 30.7

Orthopedic Surgery

Medical

2,040

29

70

9.8

6.4 to 15.1

Other

Medical

1,888

35

54

37.3

13.4 to 57.4

Physical Medicine and Rehabilitation

Medical

1,866

17

110

31.9

19.3 to 38.3

Plastic Surgery

Medical

1,848

12

154

13.3

6.1 to 23.2

Obstetrics and Gynecology

Medical

1,672

14

119

17.5

9.3 to 40.8

Dermatology

Medical

1,605

10

160

13.9

7.9 to 22.1

Trauma Therapy

Behavioral

1,581

13

122

57.4

51.7 to 62.2

Emergency Medicine

Medical

1,247

27

46

8.9

5.8 to 12.5

Nursing

Medical

1,214

45

27

18.6

3.9 to 34.2

Psychology

Behavioral

1,000

19

53

51.6

38.7 to 58.3

Massage Therapy

Medical

811

16

51

38.4

4.0 to 58.3

Cardiology

Medical

628

16

39

12.3

8.3 to 18.6

Acupuncture

Medical

626

13

48

31.9

19.1 to 52.8

Hospice and Palliative Medicine

Medical

566

12

47

30.0

17.7 to 48.5

Case Coordinators

Medical

516

19

27

2.3

1.4 to 4.7

Take it. Specialty‑level aggregates, CC BY 4.0, attribution to Twofold Research: session-length-by-specialty-2026.csv.

Why we have this data

Twofold is an ambient AI scribe: it listens during the visit and drafts the clinical note in the clinician's format, so session length is documentation exhaust. Audio is discarded once the note is drafted and no model is trained on patient data. If you would rather spend the 54 minutes on the client than the note, that is the product.

Questions about the data: info@trytwofold.com

Data and sources

Session data: Twofold internal operational dataset — 1,116,796 live conversation captures by 6,989 clinicians, six months ending July 2026. Live captures only; completed, non‑deleted visits; durations bounded 1 minute to 4 hours; medians and interquartile ranges via percentile_cont with linear interpolation. Specialty is the clinician's self‑reported profile field. Groups under 500 sessions or 10 clinicians are suppressed. All figures are de‑identified aggregates of session start and end times; no session content was accessed.

Billing bands: AMA CPT time rules for outpatient psychotherapy (90832, 90834, 90837) and CMS Medicare Physician Fee Schedule national rates; the 53‑minute threshold is the AMA midpoint between the 45‑ and 60‑minute code targets. Therapy unit steps at 23, 38 and 53 minutes follow the CMS 8‑minute rule for time‑based therapy codes.

FAQ

Frequently asked questions

  • How long is a therapy session, really?

    Across 652,932 recorded talk‑therapy sessions, median length runs 51.2 to 57.4 minutes by profession. Mental health counseling sits at 54.6, psychotherapy 53.9, clinical social work 53.8. For most professions the middle 50 percent of sessions falls between roughly 44 and 60 minutes.

  • How long is a typical psychiatry appointment?

    Median 17.4 minutes across 226,000 sessions from 1,001 clinicians. The middle 50 percent spans 8.9 to 30.4 minutes, consistent with a mix of short medication checks and longer evaluations.

  • How long is an average doctor's visit?

    Median recorded conversation: family medicine 12.1 minutes, internal medicine 13.2, pediatrics 10.2, urgent care 5.1. These measure the talking part of the visit, not waiting, silent examination, or documentation time.

  • Why are psychiatry visits shorter than therapy sessions?

    Most outpatient psychiatry is medication management. The visit exists to reach a prescribing decision, which takes 10 to 20 minutes. In therapy the conversation is the treatment, so it fills the scheduled hour.

  • Is the 50-minute hour clinical practice or billing?

    Both stories fit these medians and quartiles. CPT 90837 starts at 53 minutes and pays 15 to 25 percent more than 90834, and ten of twelve talk‑therapy medians sit within two minutes of that line. Separating the explanations requires the minute‑level distribution, which we are publishing next, whichever way it lands.

  • What data is this based on?

    1,116,796 live recorded clinical sessions from 6,989 clinicians over the six months ending July 2026, captured by an ambient scribe during the visit. Durations bounded 1 minute to 4 hours. All figures are de‑identified aggregates of session start and end times, with no access to content. Specialty‑level aggregates are downloadable as CSV under CC BY 4.0.