Safety Behaviours in CBT: Identify and Drop Them

Safety behaviours are the things anxious clients do to stop a feared catastrophe from happening: sitting near the door, rehearsing sentences before speaking, carrying water in case they faint, checking their body for signs of illness. In CBT they matter because they stop the client finding out that the catastrophe was unlikely in the first place. The actions make sense to the client, and they are one of the main reasons anxiety persists.

This guide covers what safety behaviours in CBT are, where the idea comes from, how they appear across the main models of anxiety, how to identify them with a client, and how to plan dropping them through behavioural experiments. It is written for therapists and trainees working with anxiety disorders, whether in NHS Talking Therapies, secondary care or private practice.

What is a safety behaviour?

A safety behaviour is anything a person does, in or before a feared situation, to prevent a feared catastrophe or to lessen its impact. The concept was set out by Salkovskis (1991), who argued that anxiety can persist despite repeated contact with feared situations because the person's own protective actions stop the test from being fair.

The logic runs like this. A client with panic disorder believes their racing heart means a heart attack is coming. When it starts, they sit down and breathe slowly. No heart attack happens. The client concludes, "I sat down just in time", rather than "my heart was never in danger". The feared belief survives intact, and the safety behaviour now looks essential.

Salkovskis's point was that the non-occurrence of the catastrophe gets credited to the safety behaviour. From the client's point of view, every anxious episode ends in a narrow escape. This helps explain why repeated contact with a feared situation does not always shift the belief: the client has never been in the situation without their protection.

How safety behaviours keep anxiety going

Safety behaviours maintain anxiety in several ways, and it helps to be able to explain each one to a client.

• They prevent disconfirmation. This is the core mechanism. The feared outcome does not happen, but the client attributes that to the safety behaviour, so the threat belief does not update.

• They can increase symptoms. Holding a glass tightly to stop a tremor can make the hand shake more. Breathing deeply to avert fainting can bring on light-headedness.

• They can make the feared outcome more likely. In social anxiety, rehearsing sentences and avoiding eye contact can make someone seem distant, which may bring the cool reaction they feared.

• They keep attention on threat. Monitoring the body, the audience or the exits keeps the feared outcome in mind and makes the situation feel more dangerous.

Safety behaviours across the CBT models of anxiety

Safety behaviours appear in most disorder-specific CBT models. The form differs, but the function is the same: preventing a specific feared catastrophe.

Panic disorder

In Clark's cognitive model of panic, catastrophic misinterpretation of bodily sensations drives the panic cycle (Clark, 1986). Safety behaviours are linked to the specific catastrophe feared: a client who fears fainting may hold on to furniture or sit down; one who fears suffocation may open windows or avoid tight collars. Salkovskis, Clark and Gelder (1996) described these links between particular cognitions and particular behaviours, which is why asking "what were you afraid would happen?" is often the quickest route to the safety behaviour.

Social anxiety

In the Clark and Wells model of social anxiety, safety behaviours sit alongside self-focused attention and a distorted self-image as central maintaining processes (Clark & Wells, 1995). Common examples include rehearsing what to say, keeping answers short, avoiding eye contact, hiding the face or hands, and wearing clothes to conceal sweating. Wells and colleagues (1995) examined these in-situation safety behaviours experimentally, and dropping them is a standard part of treatment built on this model.

Health anxiety

In Warwick and Salkovskis's model of health anxiety, checking the body, searching for information, and seeking reassurance from doctors, family or the internet all work as safety behaviours (Warwick & Salkovskis, 1990). Each brings brief relief, keeps attention on bodily sensations, and leaves the underlying fear of illness unchallenged.

Covert and mental safety behaviours

Not every safety behaviour is visible. Mental rehearsal, monitoring how one comes across, counting, distraction, reassuring self-talk and trying to control thoughts can all serve the same function. These are easy to miss, so ask about them directly.

Avoidance and safety behaviours: what is the difference?

Avoidance means not entering the feared situation at all. Safety behaviours are what people do when they cannot, or choose not to, avoid. The two sit on a continuum, and some writers treat safety behaviours as a subtle form of avoidance.

In practice, the distinction matters less than the function. Any behaviour whose job is to prevent a feared catastrophe will block learning about that catastrophe, whether it keeps the client out of the situation or protects them within it.

How to identify safety behaviours with a client

Clients rarely describe their safety behaviours without prompting, because to them these actions are ordinary good sense. A few strategies help:

1. Start with the feared catastrophe. Ask what the client was afraid would happen at the worst moment of a recent episode. Then ask, "What did you do to stop that happening?"

2. Ask about the "what if you hadn't" scenario. "What do you think would have happened if you hadn't sat down?" If the answer is the feared catastrophe, you have found a safety behaviour.

3. Look at function, not form. Sitting down is not a safety behaviour in itself. It becomes one when its job is to prevent a feared collapse. The same action can be neutral for one client and maintaining for another.

4. Ask about subtle and mental behaviours. "Is there anything you do in your head to get through it?" often uncovers rehearsal, monitoring or counting.

5. Observe in session. Role-plays, in-session exposure and even the waiting room can show safety behaviours the client has not mentioned.

6. Use a diary between sessions. Recording situations, feared outcomes and what the client did helps build a full list.

The Safety Behaviours Identification & Drop Plan brings this together: the feared situation, the behaviours used, what the client fears would happen without them, what each one costs, and a plan to test dropping it.

A worked example (fictional composite)

Priya is a fictional composite, not a real client. Priya, 29, is referred with social anxiety. She dreads speaking in team meetings because she believes she will go red, stumble over her words, and that colleagues will think she is incompetent.

Asked what she does to stop this, Priya describes writing out and memorising her updates, speaking quickly to get it over with, avoiding eye contact, wearing high-necked tops to hide blushing, and replaying the meeting afterwards to check how she came across. She rates her belief that "people will think I'm incompetent" at 80%.

Together, you and Priya map these onto the Clark and Wells formulation. She can see that memorising and speaking fast might make her seem stilted, and that avoiding eye contact means she never sees colleagues' actual reactions.

How to drop safety behaviours: behavioural experiments

Dropping safety behaviours works best as a planned behavioural experiment rather than an instruction. The aim is for the client to learn something new about the feared outcome, not simply to put up with discomfort.

Share the rationale first

Draw the client's own formulation and show where the safety behaviours sit. Use their examples: "If you always sit down, how would you ever find out whether you would have fainted?" Clients are more willing to drop protection when they can see what it costs them.

Make a clear prediction

Before the experiment, write down the specific prediction ("if I don't rehearse, I will go blank and people will look at each other") and rate belief in it. Vague predictions produce vague learning.

Compare with and without

A common design, drawn from treatment built on the Clark and Wells model, is to run a situation twice: once using safety behaviours and once without. The client rates anxiety and how they think they came across each time, and where possible gets feedback from others or video. The comparison is often more convincing than any discussion.

Review what happened

Afterwards, compare the outcome with the prediction, re-rate belief, and ask what the client has learned. Plan the next step. Most clients drop safety behaviours in stages, starting with those that feel least essential.

Replace with approach, not a new safety behaviour

Watch for substitutes. A client who stops carrying water may start carrying a mint instead. Encourage behaviours that engage fully with the situation, such as looking at the audience, speaking at a normal pace, or walking without holding on.

The "judicious use" debate

Not all writers agree that every safety behaviour must go straight away. Rachman, Radomsky and Shafran (2008) argued for reconsidering the case, suggesting that some safety behaviours, used judiciously, may help a client approach situations they would otherwise avoid, especially early in treatment. The question to ask is whether the behaviour is blocking the learning your experiment is designed to produce. If it is, it needs to go at some point.

A note on safety and assessment

Not everything that looks like a safety behaviour should be dropped. Before planning an experiment, assess carefully:

• Physical health. Where a client's fears concern bodily symptoms, check that appropriate medical assessment has been done. Dropping a behaviour is only a fair test once genuine medical causes have been considered.

• Prescribed medication. Medication is not a safety behaviour to be dropped as part of an experiment. Any change to prescribed medication is a decision for the client and their prescriber.

• Genuine protective behaviours. A crisis or safety plan, safeguarding arrangements, and reasonable precautions against real risk are not safety behaviours. The concept applies to behaviours that protect against an overestimated threat.

If you are unsure whether a behaviour is maintaining anxiety or protecting the client from real risk, discuss it in supervision before planning to drop it.

Frequently asked questions

What is a safety behaviour in CBT?

A safety behaviour is an action a person takes to prevent a feared catastrophe or reduce its impact, such as sitting down to avoid fainting or rehearsing what to say. Salkovskis (1991) described how these actions stop the person learning that the catastrophe was unlikely, which keeps anxiety going.

Is avoidance a safety behaviour?

They are closely related, and some writers treat safety behaviours as a subtle form of avoidance. Avoidance keeps a person out of the feared situation; safety behaviours are used within it. Both prevent the feared belief from being tested, so formulations usually include both and treatment addresses both.

Should clients drop all their safety behaviours at once?

Usually not. Most clients drop them step by step through planned behavioural experiments, starting with those that feel least essential. Some clinicians allow limited, planned use early in treatment. What matters is that, by the end, the client has tested their fears without protection.

Where to start

Pick one anxious client on your caseload and list their safety behaviours in the next session, starting from the specific catastrophe they fear. Then map each one onto their formulation and choose one to test as a behavioural experiment. The Safety Behaviours Identification & Drop Plan structures that conversation, and the interactive model pages for panic, social anxiety and health anxiety show where safety behaviours sit in each cycle.

Want interactive, model-faithful CBT worksheets you can assign as homework? Start free →

References

Clark, D. M. (1986). A cognitive approach to panic. Behaviour Research and Therapy, 24(4), 461–470. https://doi.org/10.1016/0005-7967(86)90011-2

Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In R. G. Heimberg, M. R. Liebowitz, D. A. Hope, & F. R. Schneier (Eds.), Social phobia: Diagnosis, assessment, and treatment (pp. 69–93). Guilford Press.

Rachman, S., Radomsky, A. S., & Shafran, R. (2008). Safety behaviour: A reconsideration. Behaviour Research and Therapy, 46(2), 163–173. https://doi.org/10.1016/j.brat.2007.11.008

Salkovskis, P. M. (1991). The importance of behaviour in the maintenance of anxiety and panic: A cognitive account. Behavioural Psychotherapy, 19(1), 6–19. https://doi.org/10.1017/S0141347300011472

Salkovskis, P. M., Clark, D. M., & Gelder, M. G. (1996). Cognition-behaviour links in the persistence of panic. Behaviour Research and Therapy, 34(5–6), 453–458. https://doi.org/10.1016/0005-7967(95)00083-6

Warwick, H. M. C., & Salkovskis, P. M. (1990). Hypochondriasis. Behaviour Research and Therapy, 28(2), 105–117. https://doi.org/10.1016/0005-7967(90)90023-C

Wells, A., Clark, D. M., Salkovskis, P., Ludgate, J., Hackmann, A., & Gelder, M. (1995). Social phobia: The role of in-situation safety behaviors in maintaining anxiety and negative beliefs. Behavior Therapy, 26(1), 153–161. https://doi.org/10.1016/S0005-7894(05)80088-7

This article is for qualified therapists and trainees and is educational, not clinical advice. The tools described support therapy delivered by a trained clinician and are not a substitute for assessment, supervision or clinical judgement. The example is a fictional composite.