The evidence
How many therapy sessions actually help?
If you are funding therapy for other people, this is the decision that determines whether the programme does anything. The research on it is clearer than most people expect, and it points somewhere inconvenient.
The short answer
Reviews of therapy delivered in real services put the useful dose somewhere between 4 and 26 sessions, depending on the person and the problem. The classic dose-effect study found roughly half of patients measurably improved by session eight and roughly three quarters by session twenty-six. Whether a session runs 30 minutes or 50 has far less evidence behind it than how many sessions there are and how close together they fall.
The practical consequence for anyone funding therapy: a programme that gives a large number of people one or two sessions each is buying something that sits below the bottom of every range in the literature.
Why this question comes up at all
Organisations funding therapy almost always start from a fixed budget and a headcount, and divide one by the other. It is a reasonable instinct and it produces a specific failure: a large number of people are offered a very small amount of therapy, everyone can report that support was made available, and very little measurable change happens to anybody.
The same instinct shows up as a request to shorten sessions. Twenty minutes instead of fifty doubles the number of people you can reach for the same money. It is an appealing piece of arithmetic. It is also the wrong lever, and the reason is that the research on session length is genuinely weak, while the research on the number of sessions is not.
What the dose-response research actually found
The literature on this goes back forty years and it is unusually consistent about its overall shape. More therapy produces more improvement, with diminishing returns: the curve rises steeply early and flattens later. What has changed over time is the estimate of where the useful part of the curve sits.
Across more than 2,400 patients, roughly 50% were measurably improved by session 8 and roughly 75% by session 26.
Twenty-six studies. Optimal doses in routine settings ranged between 4 and 26 sessions (4 to 6 for low-intensity guided self-help), and weekly therapy accelerated improvement compared with less frequent schedules.
Two details in the 2020 review are worth pulling out. The optimal dose varies by setting, population and outcome measure, so anyone quoting a single universal number is overselling. And weekly therapy accelerated improvement compared with less frequent schedules, which matters if your programme is going to space sessions out to make a budget last.
The finding that looks like it says the opposite
Honesty about the contested parts is the only reason to trust anyone on the uncontested parts, so here is the result that appears to contradict everything above.
In 1,868 UK primary-care clients, the rate of reliable and clinically significant improvement did not rise with sessions attended: it ran from 88% among those who attended one session down to 62% among those who attended twelve. The authors read this as clients ending treatment once they had reached a good enough level, rather than as evidence that less therapy works better.
Read quickly, that looks like an argument for funding fewer sessions. It is not. The authors call it responsive regulation, and the explanation is straightforward: people stop coming when they have got what they came for. A one-session group is therefore full of people who improved fast, and a twelve-session group is full of people whose difficulties were more stubborn. The number of sessions is not causing the difference, it is reflecting it.
What it actually implies for a funder
That endings should be decided between a person and their therapist, based on whether the therapy has done its job. The moment a funder caps everyone at a fixed low number, that regulation stops working: the people who would have kept going are the ones who needed to.
What about session length?
We are going to be plainer here than a provider selling fifty-minute sessions has any commercial reason to be. The evidence comparing session lengths is thin and mixed. Shorter formats have been used successfully in various services. Anyone telling you that fifty minutes is proven superior to thirty is going beyond what the literature supports.
What has been tested properly is frequency, and there the answer is clear.
A 200-patient randomised trial. Twice-weekly sessions produced a significantly larger fall in depressive symptoms than weekly ones (difference in effect size d = 0.55) and halved dropout.
Halving dropout is the part worth sitting with. In any funded programme, the people who quietly stop attending are the largest single source of wasted budget, and they are usually invisible in the reporting until the programme ends.
So if you are trying to get more out of a fixed spend, the levers with evidence behind them are the number of sessions and how close together they fall. Session length is not where the leverage is.
How to translate this into a budget
- Fund a course, not a taster. Pick a number of sessions per person that sits inside the evidence range rather than below it, and accept that this means reaching fewer people properly instead of more people nominally.
- Keep sessions weekly where you can. Spacing them out to stretch a budget works against the outcome you are paying for.
- Tell people the number up front. Knowing you have a defined course changes how someone uses it. Finding out mid-way that the funding has run out is worse than never starting.
- Have a route past the cap. Anyone who needs more should be able to continue with the same therapist privately rather than being handed to a stranger at the point they were getting somewhere.
- Measure sessions delivered, not people enrolled. Enrolment is the number that flatters a programme. Delivery is the number that correlates with anybody feeling better.
How we run this in practice
On every programme we set up, the funded number of sessions is agreed at the start and stated to the people using it. Each person gets a free fifteen-minute call with the therapist they choose before any paid session, because a bad match wastes the first two sessions of any course. And when the funded sessions end, the person keeps their therapist and can continue privately.
If you are working out what to fund, that is exactly the conversation we would rather have before the programme is designed than after. It is free and takes about fifteen minutes. The three programme shapes are set out under therapy for organisations, and the governance detail is on the clinical governance page.
A note on what this page is not
This is a summary of published outcome research, written for someone making a funding decision. It is not clinical advice, and it is not a substitute for the judgement of the therapist working with a particular person. Individual people vary enormously, which is precisely why capping everybody at the same low number is the mistake it is.
If you are reading this because you personally are wondering how much therapy you need, that is a better question to ask a therapist than a webpage. You can find one and speak to them free for fifteen minutes.
Common questions
There is no single number, but the research narrows it usefully. A 2020 systematic review of 26 studies of routinely delivered psychological therapy found optimal doses in routine settings ranging between 4 and 26 sessions, with 4 to 6 for low-intensity guided self-help. The classic dose-effect study found roughly half of patients measurably improved by session 8 and roughly three quarters by session 26. A programme funding one or two sessions per person sits below the bottom of every one of those ranges.
Working out what to fund?
Tell us roughly how many people and roughly what budget, and we will tell you what a programme at that size can realistically achieve, including when the honest answer is that it cannot achieve much.
Every enquiry gets a reply from a person within 2 hours, 8am to 8pm UK time, seven days a week.
More for organisations
Crisis response for customers
Independent therapy for customers affected by a recall, a safety incident or a public failure. Set up in days, not months, and held entirely outside your systems.
Incident and change support for staff
A bounded programme after a death, a serious incident on site, a redundancy round, a restructure, or a stretch of sustained abuse absorbed by a customer-facing team.
Employer-funded therapy: how it works
The mechanics of an ongoing programme: how staff book, what you are invoiced for, what you see and what you never see.
Clinical governance
Registration, insurance, vetting, data protection and confidentiality. The answer to the first question every organisation asks us.
Everything we do for organisations
The overview page: what we offer, how quickly we can start, and who each programme is for.