How Long Therapy Takes to Work, According to the Session Numbers

Eight sessions in. Eight payments, eight Thursdays built around one hour, and if somebody asked you today whether it is working, you would say you are not sure. You know you keep going. Knowing you keep going is not the same as knowing it is doing something.

The number everyone quotes, and where it comes from

Most answers to this question trace back to one paper. In 1986, Howard, Kopta, Krause and Orlinsky published a dose-effect analysis in American Psychologist that found about half of patients measurably improved by session 8, and about three quarters by session 26. In ordinary clinics the picture is different: most people attend far fewer sessions than that and improve less often. The variation between individuals is wider than either figure.

That curve came out of a reanalysis. The authors pooled fifteen data sets covering 2,431 outpatients, some of the research going back more than thirty years, and ran a probit analysis across all of it. Nobody designed a study to answer your question. They went back through what already existed.

Different summaries of that paper quote different session by session percentages. We could not settle which series is right, so none of them is here. The two figures worth carrying are the approximate ones in the authors' own abstract: about half by session 8, about three quarters by session 26. They describe the shape of a curve rather than a threshold you pass or fail.

What the curve actually says, and what it does not

Most of the movement happens early

The shape is the finding. Gains arrive fastest at the start and then flatten, so the distance between session 1 and session 8 is larger than the distance between session 8 and session 26, which is larger again than anything after that.

Flatter is not flat, and the later stretch is often where something stops being an insight and starts being a habit. But if you are eight weeks in and waiting for one dramatic week, you may be waiting for the wrong thing.

It is a distribution, not a schedule

"Half of patients improved by session 8" is a sentence about a group. It says nothing about which half you are in, and there is no rule that puts you in the earlier one for trying harder.

Two people can start on the same day with the same therapist. One notices something by week four, the other notices nothing until month five, and both of them sit inside that average.

"Improvement" meant different things in different studies

"Measurably improved" is a low bar and a moving one. In one data set it was a therapist's judgment at the end of treatment, in another a patient's rating of their own week, in a third a researcher reading case notes afterwards. Nobody filled in a standardized questionnaire every week, the way a lot of services track outcome now. You can clear that bar and still have a bad Tuesday.

The number from real clinics looks different

Sixteen years later, Hansen, Lambert and Forman went looking for what happens outside research conditions, in Clinical Psychology: Science and Practice. Two numbers sit next to each other in that paper and they do not get along.

The first is the research consensus they describe: somewhere between 13 and 18 sessions for half of patients to improve. The second comes from a national database of more than 6,000 people treated in ordinary practice. There, the average number of sessions actually attended was under five, and roughly 20% improved.

So a typical course in that data set ended about a third of the way into what the research called a dose, and the improvement rate landed at roughly a third of what carefully run treatments reported. That is not evidence that therapy does not work. It is evidence that a great many people stop long before the research expects the effect to show, usually for reasons outside the therapy itself: money, waiting lists, a job change, a therapist moving on. The money part has its own article, on what a course of therapy costs.

That is also American data, gathered in one health care system, published in 2002. None of it transfers automatically to another country, another funding model, or the way people book appointments now.

What newer reviews found

Robinson, Delgadillo and Kellett published a systematic review in Psychotherapy Research in 2020, covering 26 studies of psychological therapy as it is delivered routinely to adults. They found replicated and consistent support for a curvilinear relationship between how long treatment lasts and how it turns out. The same shape as 1986, arrived at through different decades and different measures.

The review names its exceptions too. The authors describe the evidence as scarce and inconclusive in samples with chronic or severe mental disorders, in patients with eating disorders, and in patients with intellectual disabilities. Most of what they reviewed came from university counseling centers and outpatient clinics treating common mental health problems. If that is not your situation, the curve is not describing you.

Their range for an optimal dose in routine settings is 4 to 26 sessions. The low end of that is 4 to 6 sessions of low intensity guided self-help, which services tend to offer for mild to moderate depression and anxiety; the wider range covers higher intensity therapy with patients at mixed levels of severity. Both vary by setting, clinical population and outcome measure. A range that wide is what the evidence supports when somebody asks how many sessions they need.

A competing explanation changes what a session count means at all. Barkham and colleagues set it out in 2006 in the Journal of Consulting and Clinical Psychology as the good enough level: people stop once they have what they came for, so session numbers in any data set record when treatment stopped being needed, not how long it takes. A 2021 study in the same journal by Lee and colleagues compared the two models across more than 13,000 veterans treated for depression, and the good enough level model fit better. On that reading, your session count next to somebody else's tells you almost nothing.

What moves your own timeline

Four things come up repeatedly when researchers look at what changes one person's pace. None of them is a schedule.

What you came in with

A single event with a clear before and after is different work from a pattern that has been running since you were fourteen. Grief has a shape, and so does a specific fear. Something that has been part of how you operate for twenty years usually moves on a slower timetable, and that says something about the work rather than about you. Which of the two you are dealing with belongs in the room, not in an article.

Frequency, and what gaps do to the thread

The 2020 review found that weekly therapy appears to accelerate the rate of improvement compared with less frequent schedules. A faster rate, not a different destination.

The practical version: with three weeks between appointments, a chunk of every session goes on reconstructing what happened since the last one. Going every other week for budget or availability reasons is a reasonable choice, and the clock you are measuring against stretches with it.

What happens between the sessions

One hour a week leaves 167 others. What you do with some of them is part of the work, and between-session practice is associated with better outcomes.

What the evidence does not give you is a conversion rate. Nobody can tell you that homework saves you four sessions, and anybody offering that number is making it up. The fuller picture is in therapy homework and why it gets skipped and in what you do between sessions.

Fit with the therapist

When therapy stalls for months, the relationship is one of the first things to look at, and it rarely announces itself. It shows up as a timeline that will not move rather than as a complaint you could put into words.

Two situations are easy to confuse here. A hard stretch, the kind that produces a session that leaves you wrung out, is not the same as when the fit is the thing holding it up. The first tends to pass. The second tends to repeat.

How to tell whether your own therapy is moving

This part is yours. You cannot know where you sit on somebody's curve. You can know whether the thing you came in for is different from how it was in April.

Measure the same thing, at the same rhythm, and compare months rather than single weeks. Three markers do most of the work:

• The thing you came in for, rated the same way every week, in the same words.
• Function rather than mood: sleep, answered messages, whether you show up to the things you said yes to.
• What you do in the difficult moment now, compared with what you did three months ago.

There is a whole article on the method, how to tell whether therapy is working. The part that belongs here is the rhythm. One rating is noise. Twelve weeks of the same rating is a signal.

Mood & focus-area trends
The areas you are working on, charted across weeks, so the question stops depending on how last Thursday happened to go. Try it free in MindSync →

Then compare the right units. Memory grades a whole month on whichever week was loudest. A slow stretch you can read back beats a slow stretch you are trying to recall.

Weekly & monthly summaries
Each month pulled together from what you actually logged, which is the unit this question gets answered in. Try it free in MindSync →

MindSync does not speed up therapy. It shows you the shape of it. The pace belongs to you, your therapist and whatever you walked in with, and an app gets no vote in that. What it can do is give you something better than a vague feeling to judge the pace against, which is the idea behind tracking whether therapy is working over time.

When it is fair to ask about the plan

Whenever you want. It is a normal question and it is not a complaint.

Somewhere around session 6 to 8 is a natural moment, because by then you have enough experience of the work to know what to ask. What are we working on, how will we know it is working, and roughly what shape do you expect this to take.

Most therapists treat that as part of the job. Some answer with a plan, some answer that it is too early to say and explain why, and both are real answers. A reply that stays vague across several attempts is information too. What you do with it is a decision for you and your therapist, not for an article.

Judging the pace with something other than patience

• Goals - What you came to work on, set down once and then checked against what your sessions and check-ins actually contain.
• Therapy timeline - A scrollable record of what you have really done, which is a better witness than memory when you are judging a slow month.
• Therapy relationship check - A short, regular read on the working relationship, which is the easiest thing on this list to leave unexamined for a year.

Frequently asked questions

How long does therapy take to work?

There is no single answer, but two reference points help. The 1986 dose-effect analysis by Howard and colleagues found about half of patients measurably improved by session 8 and about three quarters by session 26. Clinic data shows far fewer sessions attended and much lower improvement rates. Your own pace depends on what you came in with, how often you go, and the fit.

How many therapy sessions do I need?

Nobody can give you a number in advance, and a number tied to a specific problem would be a clinical judgment rather than a fact. A 2020 systematic review in Psychotherapy Research put the optimal dose in routine settings between 4 and 26 sessions, varying by setting, population and outcome measure. The better question is what you and your therapist are measuring.

Is it normal to feel no change after 6 sessions?

Yes, it is common. In the curve most often quoted on this, roughly half of patients had not yet shown measurable improvement by session 8, so plenty of people feel nothing definite at six. Say it out loud in the room anyway. Bring it to your therapist as information about the work rather than a verdict on it.

How long should you stay in therapy?

That is a decision for you and your therapist, and the research does not set an end date. The dose-response studies keep finding the same curve: gains come faster early and slow down later, which is a reason to review the plan periodically rather than stop on a schedule. Cost, capacity and what you are still working on all belong in that conversation.

Why does therapy take so long?

Usually because what people bring has been running for years, and because change shows up in ordinary behavior before it shows up in how you feel. A session is also one hour in a week of 168. Gaps between appointments, a poor fit and stopping early all stretch the timeline, and the last one is the most common in real clinic data.