Guide

CBT for anxiety

The most studied psychotherapy there is, and the most commonly misdelivered. Real CBT is a structured course with experiments you run between sessions. Talking about your thoughts for fifty minutes isn’t it.

Reviewed by Albert Wong, PhDClinical psychologistLast reviewed September 2026About this site

Cognitive behavioural therapy for anxiety is a structured, time-limited course — typically eight to twenty sessions — with an agenda each week, measurable targets, and work to do between appointments. That structure isn’t bureaucratic. It’s what distinguishes it from a supportive conversation, and it’s the part most often dropped.

What actually happens

Formulation first. You map your own loop: the trigger, the thought, the feeling, the physical response, and what you do about it. Seeing it drawn out is frequently the first useful thing, because the maintaining behaviour becomes visible.

Then the thinking. Identifying the prediction — not vague worry, but the specific forecast: I will freeze, they will notice, I will be humiliated. Then examining it: what is the actual evidence, what has actually happened before, what would you tell someone else.

Then the behaviour, which is the real work. Behavioural experiments: going and finding out. Predict what will happen, do the thing, record what actually happened. This is where the change comes from, and it’s why CBT without homework is essentially inert.

And dismantling safety behaviours. The drink before the party, the seat by the door, the over-preparation, the phone in your hand. These make each occasion survivable and keep the fear intact, because your brain concludes the precaution was why nothing went wrong.

What the evidence says

CBT has the strongest evidence base of any psychotherapy for anxiety disorders, with large effect sizes across panic, social anxiety, GAD and specific phobia, and gains that hold at follow-up better than medication alone. It’s first-line in essentially every clinical guideline.

Two caveats worth stating. The effects are strongest for panic and phobia, more modest for GAD. And “CBT” in a trial means a manualised protocol delivered with fidelity, which isn’t always what is being delivered in the room.

How to tell whether you’re getting real CBT

There’s an agenda each session, agreed at the start. There’s homework, and it’s reviewed. There are measures — a questionnaire score tracked over time. There’s a number of sessions in mind. And there are behavioural experiments, not only conversations about thoughts.

If six months have passed with none of that, you may be receiving supportive therapy with cognitive vocabulary. That can be valuable; it isn’t the treatment with the evidence.

Where it does less well

Standard cognitive work is a poor fit for OCD, where examining the thought is a compulsion. It’s also a weaker starting point where the main difficulty is the struggle against the feeling itself rather than the content of the thought — that’s where ACT often lands better.

And for anyone whose experience has been of being told their reactions are wrong, the challenging can land as one more person saying so. A good CBT therapist notices that and adjusts; a mechanical one doesn’t.

Keep reading

47 guides on anxiety and what treats it, free and without an account. There is a directory of therapists here too — it is new, and growing.

General information, not clinical advice; the people described are composites, not real clients. New or sudden physical symptoms deserve a medical opinion before they are called anxiety. Listings on this site are maintained by independent practices. If you are in immediate danger, call 911; the 988 Suicide & Crisis Lifeline is available by call or text any time.