Does self-hypnosis work? Where the evidence is strongest
Self-hypnosis can work for some goals, and the evidence is strongest for anxiety and for the anxiety and pain that come with medical procedures. A 2024 umbrella review brought together 49 meta-analyses covering 261 studies. It found the most robust evidence for people undergoing medical procedures (12 reviews, 79 studies) and for people with pain (4 reviews, 65 studies).
The same review has caveats: only 9 of the 49 meta-analyses were rated high quality, and fewer than half reported any safety data.
A 2025 meta-analytic update points the same way: across 20 randomized trials with 1,250 patients having invasive procedures, hypnosis reduced anxiety and pain compared with standard care, with minimal side effects. Much of this research took place in hospitals and clinics. It shows that hypnosis can help; it can't promise that a particular session will help you, and results vary. If you have a procedure or a big day coming up, see self-hypnosis before a stressful event.
Does self-hypnosis work for anxiety?
It can help, and it seems to work best as part of wider care. A 2019 meta-analysis of 17 controlled trials found that by the end of treatment, the average person who received hypnosis had reduced their anxiety more than about 79% of people in control groups. In the 7 trials that followed people for longer, the figure was about 84%. Hypnosis was more effective combined with other psychological approaches than on its own.
An expert summary from the International Society of Hypnosis (2026) agrees. It places the strongest evidence on procedural, medical and performance anxiety, and calls the evidence for chronic anxiety disorders “promising but less extensive.” It notes that psychological treatments such as cognitive behavioral therapy are the first-line options for anxiety disorders, and that hypnosis is “unlikely to represent a universal or sufficient treatment on its own.”
So use self-hypnosis for anxiety to practice a calmer first response to the situations that set it off, alongside professional care when anxiety shapes your days. For looping thoughts, see self-hypnosis for overthinking and racing thoughts.
Does self-hypnosis work for stress, sleep and focus?
It may, but the evidence is thinner and less certain.
Everyday stress. A 2017 review of 9 small trials (365 healthy people, mostly medical students) found less stress with hypnosis in 6, but all had a high risk of bias, so the authors called the effect “still unclear.” In a 2020 trial of 95 stressed adults, five weekly group sessions plus home recordings lowered stress more than a booklet alone (41.8 versus 62.9 out of 100 after five weeks), and stress was still lower at 12 weeks.
A single session. A 2026 randomized study of 80 adults (a preprint, not yet peer-reviewed) found no evidence of lasting change in daily mood, stress, worry or sleep in the week after one 30-minute session. Its authors note that many studies with benefits used several sessions over weeks. So how long does self-hypnosis take to work? There's no fixed timeline: think in weeks of practice, not one sitting. Our FAQ covers how often to listen.
Sleep. For trouble sleeping at night, research on sleep hypnosis is limited: in a 2018 review of 24 studies, 14 reported a benefit, 3 mixed results and 7 none, with small samples and low quality overall.
Focus. We found no trial testing whether self-hypnosis improves everyday focus. If you use it for focus, treat it as a way to rehearse the steady attention you want, not a proven fix.
Is hypnosis real, or is it just relaxation?
Hypnosis is real: a measurable state of focused attention, not a stage trick and not simply relaxation. An expert summary from the International Society of Hypnosis describes it as involving “focused attention, absorption, and increased responsiveness to therapeutic suggestions.” It adds that hypnosis “does not involve loss of control” and is usually experienced as a collaboration: people remain aware and able to reject suggestions that don't fit their values or goals.
Brain imaging backs this up. Stanford researchers screened 545 healthy adults and scanned 57 at the two extremes of hypnotizability while they rested, recalled a memory and followed pre-recorded hypnosis instructions. They found three changes in brain activity and connectivity, which the authors link to focused attention, control over body and emotions, and less self-consciousness. The changes appeared only in the highly hypnotizable group, and only during hypnosis.
Relaxation is often part of a session, but the defining features are absorbed attention and response to suggestion. Self-hypnosis is hypnosis you practice on your own, often by following a recorded voice. For how it differs from meditation, see self-hypnosis vs. meditation.
Do self-hypnosis recordings work as well as a live hypnotherapist?
In the research available, recordings hold up well. A 2002 meta-analysis of 20 controlled studies with surgical patients found that people who received hypnosis did better than 89% of patients in control groups, with no significant difference between live and recorded hypnosis.
Recordings have also helped on their own. In a 2013 randomized trial with 35 adults, a hypnosis audio recording used for two weeks had a medium to large effect on stress, burnout and well-being compared with a wait-list.
A live hypnotherapist can adapt to you in the moment, and that matters most for complex problems. The International Society of Hypnosis notes that clinicians need appropriate training to use hypnosis safely, particularly with trauma-related anxiety or complex psychiatric presentations. For everyday stress, a well-made guided hypnosis recording or hypnosis app is a reasonable place to start. For trauma or a diagnosed condition, look for a clinician trained in hypnotherapy near you.
Our own sessions are recordings. Mind Refuge (mindrefuge.ca) builds a personalized 25-minute session from recordings by certified hypnotherapist David Boulet; every word is his real voice, not AI. None of the studies in this article tested Mind Refuge.
Can anyone be hypnotized?
Most people respond to hypnosis to some degree, but how strongly varies a lot. Only about 10% of people are generally categorized as highly hypnotizable, according to Stanford Medicine.
That range shows up in results. In a 2019 meta-analysis of 85 lab studies with 3,632 healthy volunteers, hypnosis with direct pain-relief suggestions reduced pain by 42% in highly suggestible people and by 29% in medium responders, while low responders gained little. The authors suggest that hypnosis “can deliver meaningful pain relief for most people.” The International Society of Hypnosis summary notes that hypnosis appears particularly useful for highly responsive people, but that beneficial effects “are not limited to highly hypnotizable patients.”
Can self-hypnosis work if you don't believe in it? Possibly, but expectations seem to play a part. In a trial with 200 women having breast surgery, what patients expected to feel explained part of the effect of hypnosis on their pain and fatigue afterward. It's worth going in curious rather than braced against it, giving it a few weeks, and stopping if it doesn't help.
What can't self-hypnosis do?
Self-hypnosis can't stand in for medical or psychological care, and it doesn't work the same way for everyone. Based on the studies above:
- It isn't a stand-alone treatment for an anxiety disorder. Psychological treatments such as cognitive behavioral therapy come first; hypnosis is often studied and used as an add-on.
- A single session may not be enough. In the 2026 study, one 30-minute session produced no detectable change in daily mood, stress or sleep over the following week.
- It doesn't take over your mind. You remain aware and able to reject suggestions.
- It isn't established for focus or sleep problems. Studies there are few, small or missing.
- It isn't for every moment. Never listen while driving or doing anything that needs your attention. With trauma-related anxiety or a complex mental health condition, work with a trained clinician rather than on your own.
What the research says, study by study
These studies looked at hypnosis in general, not at Mind Refuge. Our research page gathers more studies by topic.
Rosendahl, Alldredge & Haddenhorst (2024): 49 meta-analyses
- Design: umbrella review of 49 meta-analyses of randomized trials, covering 261 studies (Frontiers in Psychology).
- Found: the most robust evidence for medical procedures and pain.
- Limit: only 9 were rated high quality; fewer than half reported safety data.
Walter et al. (2025): invasive medical procedures
- Design: meta-analysis of 20 randomized trials with 1,250 patients (PubMed).
- Found: less anxiety and pain than with standard care, and minimal side effects.
- Limit: techniques and outcome measures varied; clinical settings only.
Valentine et al. (2019): anxiety
- Design: meta-analysis of 17 controlled trials (PubMed).
- Found: the average person given hypnosis reduced anxiety more than about 79% of controls, more so alongside other psychological care.
- Limit: a small evidence base; only 7 trials reported longer-term results.
Fisch, Brinkhaus & Teut (2017): stress
- Design: systematic review of 9 randomized trials with 365 healthy participants, mostly medical students (PubMed).
- Found: 6 of the 9 reported less stress with hypnosis.
- Limit: all at high risk of bias; the authors called the effect “still unclear.”
Fisch et al. (2020): group hypnosis for stress
- Design: randomized trial with 95 healthy but stressed adults: 5 weekly group sessions plus 5 home recordings, versus a booklet alone (PubMed).
- Found: lower stress at 5 weeks (41.8 versus 62.9 out of 100) and at 12 weeks.
- Limit: the comparison was a booklet, not another active practice.
Wesarg-Menzel et al. (2026): a single session
- Design: randomized trial with 80 adults not selected for high stress: one 30-minute session with a safety cue, versus a control visit, and a week of phone check-ins before and after (Research Square).
- Found: no evidence of lasting change in mood, stress, worry, rumination or sleep quality; an exploratory trend toward lower heart rate before bed.
- Limit: not yet peer-reviewed; one session; participants knew their group.
Chamine, Atchley & Oken (2018): sleep
- Design: systematic review of 24 studies in adults (PubMed).
- Found: 14 reported a sleep benefit, 3 mixed results and 7 none; few adverse events.
- Limit: few studies of people with sleep complaints, small samples, low quality.
Jiang et al. (2017): hypnosis in the brain
- Design: brain imaging of 57 healthy adults chosen from 545 for very high or very low hypnotizability, using pre-recorded hypnosis instructions (PubMed; Stanford Medicine).
- Found: three changes in brain activity and connectivity, only in the highly hypnotizable group during hypnosis.
- Limit: it describes the state; it tests no health benefit.
Montgomery et al. (2002): live versus recorded
- Design: meta-analysis of 20 controlled studies with surgical patients (PubMed).
- Found: patients given hypnosis did better than 89% of controls, with no significant difference between live and audiotaped hypnosis.
- Limit: older studies, all in surgery.
Cardeña, Svensson & Hejdström (2013): a recording for stress
- Design: randomized trial with 35 adults: a hypnosis audio recording used for two weeks, versus a wait-list (PubMed).
- Found: medium to large improvements in stress, burnout and well-being.
- Limit: a small sample, and the comparison group received nothing.
Thompson et al. (2019): who responds
- Design: meta-analysis of 85 lab studies with 3,632 healthy volunteers and experimentally induced pain (PubMed).
- Found: with direct pain-relief suggestions, pain fell 42% in high and 29% in medium responders; low responders gained little.
- Limit: lab pain in healthy people; clinical data are still needed.
Montgomery et al. (2010): expectations
- Design: randomized trial with 200 women having breast-conserving surgery: brief hypnosis before surgery versus an attention control (PubMed).
- Found: what patients expected explained part of the effect of hypnosis on pain and fatigue afterward.
- Limit: one surgical setting; it doesn't test skeptics directly.