what workslast verified 2026-09-23

how long should therapy take?

there is a version of therapy that is a standing appointment for life, and a culture that treats it as a sign of self-awareness. the research does not describe that. it describes a treatment with a dose, a curve, and an end: most of the change happens in the first handful of sessions, most people leave when they have what they came for, and the people who stay for years are, on average, the people who came in worst and changed slowest. here is the record, the strongest case for the other side, and what to ask.

five questions, answered from the data

how many sessions do most people need?[1][2]

between 4 and 26 in routine services, 4 to 6 for guided self-help, across 26 studies. in the biggest single data set, 102,206 NHS patients, 95 percent of the people who were going to get better had done so within 7 sessions of low-intensity work or 14 of full CBT. social anxiety, PTSD and OCD ran longer, 6 to 16.

does more therapy mean more improvement?[3][4][5]

not on average. in 1,868 primary-care clients with planned endings, the share who improved fell from 88 percent at one session to 62 percent at twelve. in 114,123 patients across 15 studies there was no overall link between how long treatment ran and how people ended up. the people who stay longer came in worse and change more slowly; they are not getting more out of each week.

so does length not matter at all?[6][7][8]

it matters less than frequency. in 70 depression trials the number of sessions had no independent effect once other things were held constant, but going from one session a week to two, with the same total, raised the effect by g 0.45. weekly beats fortnightly in 21,488 counselling clients; twice-weekly beat weekly in a 200-person randomised trial. if you can only afford ten sessions, take them weekly, not monthly.

what is the case for years of therapy?[9][10][11]

a 2008 JAMA meta-analysis of 23 studies reported that psychodynamic therapy lasting a year or more beat shorter therapy in complex disorders by a very large margin (effect size 1.8). a 2010 re-analysis found a miscalculation in the effect sizes, small heterogeneous studies, and comparisons that included no therapy at all, and concluded there was no evidence of superiority. the one randomised trial that ran five years found short-term therapy better in the first year and long-term therapy ahead on two measures later. that is the strongest case, and it is a mixed one.

when should therapy end?[2][3][4]

the field’s own term is the good-enough level: people stop when they have what they came for, and that is why the aggregate curve flattens. the data-set authors recommend against a fixed limit for everyone, and for a progress review at the point where most responders have responded: around session 7 for light-touch work and 14 for full therapy. if there is no measured change by then, the question is not “more of this” but “what else.”

the shape of the curve, and what it hides

plotted across everyone, improvement against sessions is a curve that rises fast and flattens: the famous dose-response curve.[1] the good-enough-level studies showed why it flattens. people do not all improve on the same slow schedule; the ones who get what they need early leave, and the people still in the room at session twenty are the ones for whom it is slower going.[3][4] so the flat part of the curve is not evidence that late sessions do little for the people who need them. it is evidence that most people did not need them. both of those sentences are true, and a page that keeps only one is selling either endless therapy or none.

the other direction

the review that gives the 4-to-26 range says plainly that the evidence for chronic and severe disorders is scarce and inconclusive.[1] the long-term psychodynamic meta-analysis, whatever its arithmetic, was about complex conditions: personality disorders, chronic and multiple diagnoses.[9] the finnish trial found long-term therapy ahead on sense of coherence and perceived competence years later, after being behind in year one.[11] if that is your situation, the short-course numbers on this page were not measured on you. the question to bring is still the same one: what is being measured, and is it moving.

what to ask

questions people ask

How long does therapy usually take?

Most people who improve do so within 4 to 26 sessions in routine services, and within 4 to 6 sessions of guided self-help. In a data set of 102,206 UK patients, 95 percent of responders had improved within 7 sessions of low-intensity CBT or 14 sessions of high-intensity CBT. Social anxiety, PTSD and OCD tended to need the longer end.

Is it bad to be in therapy for years?

It is not what the research describes as typical or necessary for most problems. Longer treatments in the data go to people who started with more severe symptoms and change more slowly; there is no overall association between duration and outcome across 114,123 patients. For complex conditions there is a contested case that long-term psychodynamic therapy helps more, and one five-year randomised trial with a split result. If you have been in therapy for years, the useful question is whether anything measured has changed.

Is twice a week better than once a week?

For depression, on average, yes. A metaregression of 70 trials found that doubling the weekly frequency with the same total number of sessions raised the effect size by 0.45; a 200-person randomised trial found twice-weekly sessions gave better outcomes than weekly. In 21,488 counselling clients, weekly sessions produced faster recovery than fortnightly. Frequency matters more than total length.

Does this page say anything about medication?

No. It is about the length and frequency of psychological therapy only. Nothing here is a reason to change any treatment on your own.

sources

  1. Robinson L, Delgadillo J, Kellett S. The dose-response effect in routinely delivered psychological therapies: a systematic review. Psychotherapy Research 2020;30(1). 26 studies. “Replicated and consistent support” for a curvilinear relationship between length and outcome; “optimal doses of psychotherapy in routine settings range between 4 and 26 sessions (4-6 for low intensity guided self-help)”; “weekly therapy appears to accelerate the rate of improvement compared to less frequent schedules”; evidence “scarce and inconclusive” for chronic and severe disorders. PMID 30661486. https://doi.org/10.1080/10503307.2019.1566676
  2. Robinson L, Kellett S, Delgadillo J. Dose-response patterns in low and high intensity cognitive behavioral therapy for common mental health problems. Depression and Anxiety 2020;37:285-294. 102,206 patients across 16 NHS services. “Most responders (95%) attained [reliable and clinically significant improvement] within 7 sessions of LiCBT and 14 sessions of HiCBT”; social anxiety, PTSD and OCD needed high-intensity work and 6 to 16 sessions. PMID 32027435. https://doi.org/10.1002/da.22999
  3. Barkham M, Connell J, Stiles WB, et al. Dose-effect relations and responsive regulation of treatment duration: the good enough level. Journal of Consulting and Clinical Psychology 2006;74:160-167. 1,868 primary-care clients with planned endings, 1 to 12 sessions. Reliable and clinically significant improvement “did not increase with number of sessions attended”: 88 percent of those who attended one session, down to 62 percent at twelve (r = −0.91). The aggregate curve “may reflect progressive ending of treatment by clients who had achieved a good enough level of improvement.” PMID 16551153. https://doi.org/10.1037/0022-006X.74.1.160
  4. Baldwin SA, Berkeljon A, Atkins DC, Olsen JA, Nielsen SL. Rates of change in naturalistic psychotherapy: contrasting dose-effect and good-enough level models of change. Journal of Consulting and Clinical Psychology 2009;77:203-211. 4,676 patients, median 5 sessions. “Small doses of treatment were related to relatively fast rates of change, whereas large doses of treatment were related to slower rates of change”; uniform time limits “would not adequately serve patients’ needs.” PMID 19309180. https://doi.org/10.1037/a0015235
  5. Bone C, Delgadillo J, Barkham M. A systematic review and meta-analysis of the good-enough level (GEL) literature. Journal of Counseling Psychology 2021;68:219-231. 15 studies, n = 114,123. “No overall association between treatment duration and outcomes (r = −0.24, p = .27)”; longer treatments had higher baseline symptoms (r = 0.15) and slower change; “most people nonetheless responded within defined boundaries.” PMID 33090874. https://doi.org/10.1037/cou0000521
  6. Cuijpers P, Huibers M, Ebert DD, Koole SL, Andersson G. How much psychotherapy is needed to treat depression? A metaregression analysis. Journal of Affective Disorders 2013;149:1-13. 70 studies, 5,403 patients. Number of sessions: “only a small association… no longer significant” after adjustment; no association with total contact time or duration; “an increase from one to two sessions per week increased the effect size with g = 0.45, while keeping the total number of treatment sessions constant.” PMID 23528438. https://doi.org/10.1016/j.jad.2013.02.030
  7. Erekson DM, Lambert MJ, Eggett DL. The relationship between session frequency and psychotherapy outcome in a naturalistic setting. Journal of Consulting and Clinical Psychology 2015;83:1097-1107. 21,488 university counselling clients over 17 years. Weekly attendance produced “steeper recovery curves” and faster clinically significant gains than fortnightly; “few significant differences… in total amount of change.” PMID 26436645. https://doi.org/10.1037/a0039774
  8. Bruijniks SJE, et al. Individual differences in response to once versus twice weekly sessions of CBT and IPT for depression. Journal of Consulting and Clinical Psychology 2022;90(1). Reports the parent trial (n = 200, weekly vs twice-weekly): “on average, twice-weekly sessions of psychotherapy for depression lead to better outcomes compared to once-weekly sessions,” with subgroups differing (d = 0.37 for matched vs unmatched frequency). PMID 35225634. https://doi.org/10.1037/ccp0000658
  9. Leichsenring F, Rabung S. Effectiveness of long-term psychodynamic psychotherapy: a meta-analysis. JAMA 2008;300:1551-1565. 23 studies, 1,053 patients, therapies of at least a year or 50 sessions, in complex disorders. Reported LTPP superior to shorter therapies with a between-group effect size of 1.8 (95% CI 0.7 to 3.4). PMID 18827212. https://doi.org/10.1001/jama.300.13.1551
  10. Bhar SS, Thombs BD, Pignotti M, et al. Is longer-term psychodynamic psychotherapy more effective than shorter-term therapies? Review and critique of the evidence. Psychotherapy and Psychosomatics 2010;79:208-216. Re-examined the JAMA meta-analysis: “a miscalculation of the effect sizes used to make key comparisons”; claims “depended on a set of small, underpowered studies that were highly heterogeneous”; LTPP compared against 12 kinds of control including no therapy; “we found no evidence to support claims of superiority of LTPP over shorter-term methods.” PMID 20424498. https://doi.org/10.1159/000313689
  11. Knekt P, Lindfors O, Sares-Jäske L, Virtala E, Härkänen T. Randomized trial on the effectiveness of long- and short-term psychotherapy on psychosocial functioning and quality of life during a 5-year follow-up. Psychiatry Research 2015;229(1-2). Helsinki Psychotherapy Study, 326 outpatients with mood or anxiety disorders randomised to solution-focused, short-term psychodynamic, or long-term psychodynamic therapy, followed five years. “Short-term therapies improved psychosocial functioning and quality of life more than LPP during the first year”; later, sense of coherence and perceived competence improved more with long-term therapy; “short-term therapy has consistently more short-term effects… whereas LPP has some additional long-term benefits.” PMID 26162657. https://doi.org/10.1016/j.psychres.2015.05.113
  12. Howard KI, Kopta SM, Krause MS, Orlinsky DE. The dose-effect relationship in psychotherapy. American Psychologist 1986;41:159-164. The paper that started the field; its figures are widely quoted but its abstract is not on PubMed and it was not read for this page, so no number from it is used here. PMID 3516036. https://pubmed.ncbi.nlm.nih.gov/3516036/

related: what works that costs nothing · who wrote the PHQ-9

adam