Anxiety: What Research Says About CBT, Exercise, Mindfulness and Breathing

Anxiety disorders affect roughly 1 in 5 US adults in a given year, according to NIMH figures from a 2001–2003 survey. Cognitive behavioral therapy has the strongest evidence, exercise and mindfulness show promise, and breathing techniques have the weakest. This article lays out how large the effects are and where the evidence is thin.

Anxiety is common enough that most adults will recognize it, and anxiety disorders are common enough to count as a public health issue. According to the National Institute of Mental Health (NIMH), roughly 1 in 5 US adults had an anxiety disorder in the past year, with women affected more often than men. The research on what helps does not point to a single cure. It points to several approaches, each with a moderate average effect and wide variation between individuals.

That matters because advice about anxiety often blurs ordinary worry with clinical disorders. Most of the trials discussed below enrolled people with diagnosed anxiety disorders or noticeable symptoms, not people coping with everyday stress, so their results may not transfer cleanly in either direction. What follows sorts the approaches by how well they have been tested, how large the measured effects are, and how much confidence the evidence deserves.

How common anxiety is, and where worry ends and a disorder begins

The national figures come with a large caveat: they are old. NIMH's adult estimates derive from the National Comorbidity Survey Replication, a diagnostic-interview survey conducted in 2001–2003 using the definitions of that era (DSM-IV). A secondary site, mhstats.org, reports that these remain the most recent nationally representative estimates NIMH publishes, though that claim could not be verified on NIMH's own pages. The numbers may not capture changes since 2020, and no more recent national data were available for this article.

Within that survey, mhstats.org, citing NIMH, reports a past-year prevalence of 19.1% and a lifetime prevalence of 31.1%. An older NIMH fact sheet gives a 12-month figure of 18.1% and classifies 22.8% of cases, equal to 4.1% of all adults, as severe. The gap between 18.1% and 19.1% probably reflects two analyses of the same survey, although that could not be confirmed, so "roughly 1 in 5" is the safest summary.

Differences by sex and by disorder

NIMH reports that the past-year prevalence of any anxiety disorder was 23.4% in women and 14.3% in men. For social anxiety disorder, NIMH puts the past-year figure at 7.1% of adults, and 29.9% of those cases were seriously impaired. Generalized anxiety disorder (GAD), characterized by persistent worry, had a past-year prevalence of 3.4% in females and 1.9% in males.

The National Alliance on Mental Illness (NAMI), an advocacy organization that cites the same NIMH figures, adds that 11% of youth aged 3–17 have anxiety problems and that most people develop symptoms before age 21. Because these sources all trace back to one survey, they are not independent confirmation of one another.

Reading the effect sizes

The studies below report results as standardized effect sizes, labeled SMD, Hedges' g or ES. By a general statistical convention, not one drawn from the studies themselves, about 0.2 is considered small, 0.5 moderate and 0.8 large. In the exercise studies, a negative SMD means lower anxiety than in the comparison group. Because the studies measure different outcomes in different populations, the numbers cannot be lined up as a ranking.

Cognitive behavioral therapy: the best-tested option

CBT has the longest and broadest evidence base of any approach considered here, and most anxiety-disorder trials involve it. A 2018 meta-analysis by Carpenter and colleagues in Depression and Anxiety pooled randomized, placebo-controlled trials of CBT for anxiety and related disorders. It found a moderate effect on the symptoms of the targeted disorder (Hedges' g = 0.56) and smaller effects on other anxiety symptoms (g = 0.38), depression (g = 0.31) and quality of life (g = 0.30). The odds of responding to treatment were 2.97 times those of placebo.

The authors described CBT as "moderately efficacious" against placebo, and the wording is deliberate. They also said better treatments are especially needed for PTSD, social anxiety disorder and panic disorder. In the PTSD trials, dropout was higher with CBT (29.0%) than with placebo (17.2%), a reminder that therapy is demanding as well as effective.

An average, not a guarantee

The pooled figure is a group average. Individuals vary widely, and some people do not respond. The placebo-controlled effect of about 0.56 is also lower than what older meta-analyses reported using waitlist controls, a design that tends to inflate effects because people waiting for treatment have little reason to improve.

A meta-analysis of CBT's effect on quality of life, covering 44 studies, 59 trials and 3,326 participants, found moderate effects in both controlled and pre-post comparisons (g = 0.56 and 0.54). Effects declined with later publication year, a possible sign that newer trials show smaller benefits. Separately, a meta-analysis of 56 real-world effectiveness studies found large pre-to-post improvements in disorder-specific symptoms, in the range seen in controlled trials, although only six of those studies had comparison groups.

Lighter-touch and online formats

Not everyone can see a therapist weekly, so researchers have tested lower-intensity versions: self-help materials, guided self-help and psychoeducational groups. A meta-analysis of low-intensity CBT for GAD pooled 12 effect sizes, with 1,201 participants contributing anxiety outcomes. Results varied significantly across studies for both anxiety and worry. The authors cite other work suggesting guided self-help tends to outperform pure self-help, but the effect sizes were not available for this article.

On delivery format, the quality-of-life meta-analysis found face-to-face treatment outperformed internet-delivered treatment. A 2022 meta-analysis compared the two formats for GAD, but its results could not be reviewed here. No conclusions about app-based CBT can be drawn from the evidence gathered.

Exercise: consistent direction, uncertain size

Reviews of exercise for anxiety mostly point the same way: people who exercise report lower symptoms than comparison groups. A 2026 systematic review and meta-analysis in BMC Sports Science, Medicine and Rehabilitation pooled 10 studies with 2,408 adults with generalized anxiety symptoms and found an SMD of −0.502 (95% confidence interval −0.830 to −0.174). That is a moderate average effect, with a wide range of plausible values.

The authors themselves called the evidence only "suggestive," citing exceptionally high heterogeneity and limited blinding, and said it should be read with low confidence. They also reported a larger apparent benefit in middle-aged and older adults (SMD −1.063) than in younger adults (−0.388), which they described as a preliminary subgroup finding rather than an established difference.

Other age groups and types of exercise

A meta-analysis of 11 randomized trials with 770 older adults found an SMD of −0.60 (95% CI −0.88 to −0.32). Its authors called the evidence "strong," which is more confident wording than a pool of 11 small trials comfortably supports. For children and adolescents, a review of 22 studies (39 effects) spanning 1982 to 2023 found an effect size of 0.71 (95% CI 0.36 to 1.06), with moderate and high intensities working better.

On which kind of exercise works best, a network meta-analysis of 23 randomized trials with 1,830 participants concluded that all eight modalities it examined improved anxiety and depressive symptoms. They were yoga, Tai Chi, Pilates, resistance training, aerobic exercise, combined programs, high-intensity interval training and moderate-intensity continuous training. Effect sizes and any ranking were not available for this article, so no modality can be called the best.

Several limits apply across these reviews. They rely on small trials and subjective symptom ratings, participants cannot be blinded to whether they are exercising, and the reviews often combine different populations and anxiety scales. The reasonable reading is that exercise likely lowers anxiety modestly and carries low risk, while its size and best form remain unsettled.

Mindfulness and medication: one trial, carefully read

The most discussed comparison between a non-drug approach and medication is a single randomized trial by Hoge and colleagues, published in JAMA Psychiatry in 2023 (online November 2022). It randomized 276 adults with anxiety disorders: 136 to an 8-week mindfulness-based stress reduction (MBSR) course and 140 to the medication escitalopram. The primary analysis covered 208 completers (102 MBSR, 106 escitalopram), with a mean age of 33 and 75% women.

MBSR was noninferior to escitalopram on the main clinician-rated measure, the Clinical Global Impression of Severity (CGI-S), which fell 1.35 points with MBSR and 1.43 with escitalopram. A 2024 secondary analysis in JAMA Network Open found that a few measures favored escitalopram at midtreatment, but not by the end of the study. Noninferiority means MBSR performed about as well as the drug within a preset margin. It does not mean MBSR is better.

Adverse events and dropout

According to the secondary analysis, at least one study-related adverse event occurred in 78.6% of participants taking escitalopram (110 people) and 15.4% of those in MBSR (21 people). The trial's published abstract adds that 8% of the escitalopram group (10 people) left the study because of adverse events, versus none in the MBSR group. The comparison is partly apples to oranges, because adverse events were collected differently for a drug and for a behavioral class.

What the trial required and what it cannot show

MBSR was not casual self-help. The registered protocol on ClinicalTrials.gov calls for an instructor-led course plus 45 minutes of daily home practice. Some secondary coverage described it as a "self-help intervention," which misstates the design.

The trial had no placebo or no-treatment arm, so it cannot show that MBSR beats a placebo. Participants knew which treatment they received, which leaves room for expectation effects, and the sample skewed female. Multiple news and web accounts of the trial all rest on this one study and do not count as independent confirmation. The funding and conflict-of-interest statements could not be checked, and the finding needs replication.

Slow breathing: plausible, low-risk, uncertain

Breathing techniques are the easiest of these approaches to try, and the evidence for them is the least settled. A News-Medical summary of a 2023 Scientific Reports meta-analysis of breathwork trials reports lower stress than in control groups and a high safety profile for slow-paced breathing. Only the summary was reviewed, not the paper itself.

A 2025 review of pranayama, a yogic breathing practice, cites a 26-trial meta-analysis that found small-to-medium effects of breathwork on stress, anxiety and depression. According to that description, only slow breathing produced significant pooled effects. This too is a secondary account of the underlying analysis.

The counterevidence

A 2024 review by Chin and colleagues in Frontiers in Psychology examined brief, single-session interventions for state anxiety, meaning anxiety felt in the moment. Breathing-focused interventions did not significantly reduce state anxiety compared with control conditions. A 2018 systematic review of slow breathing in healthy people, which screened 2,461 abstracts and included 15 studies, likewise found mixed anxiety results, with some studies showing reductions and others not.

One way to reconcile the findings is that sustained practice over weeks may help while a single session may not, but that is an interpretation of the pattern, not something the reviews prove. Reviews also mix healthy volunteers with patients, and people cannot be blinded to a breathing technique. Slow breathing is best described as a low-risk, plausible aid with an uncertain effect size.

Screening and professional evaluation

On June 20, 2023, the US Preventive Services Task Force (USPSTF) recommended screening for anxiety disorders in adults aged 64 and younger, including pregnant and postpartum people, and gave the recommendation a Grade B. For adults 65 and older it found the evidence insufficient to recommend for or against screening, which is not the same as advising against it. The Task Force concluded with moderate certainty that screening has a moderate net benefit.

Two limits are worth keeping in view. The USPSTF found that screening tools alone cannot diagnose an anxiety disorder, so a positive screen leads to further evaluation rather than a diagnosis. It also found no evidence on the best interval between screenings. The recommendation concerns people without recognized symptoms and is not a treatment guideline. A summary in American Family Physician notes that treatment can include psychotherapy or medication but offers no comparison between them.

What the evidence adds up to

Taken together, the research supports a modest but practical message. CBT is the best-supported psychological treatment, with a moderate average effect against placebo. Exercise probably helps a little and carries low risk, an 8-week mindfulness course matched a medication in one trial, and slow breathing remains an inexpensive but unproven aid. None of these amounts to a cure, and each average conceals people who benefit a great deal and people who do not.

The gaps are as informative as the findings. Better trials of mindfulness against placebo and against CBT, newer prevalence data, and evidence on which approach suits which person would turn broad averages into usable guidance. Until then, the most reliable step the evidence supports is to treat persistent, impairing anxiety as something to have evaluated by a professional.

Frequently asked questions

What is the best-supported treatment for anxiety?

Cognitive behavioral therapy has the longest and broadest evidence base. A 2018 meta-analysis of placebo-controlled trials found a moderate effect on the target disorder's symptoms (Hedges' g = 0.56). The authors called CBT "moderately efficacious" and said better treatments are still needed for PTSD, social anxiety disorder and panic disorder.

Does exercise reduce anxiety?

Reviews generally find lower anxiety symptoms in people who exercise, with moderate average effects. A 2026 review of adults with generalized anxiety symptoms found an SMD of −0.502, but its authors rated confidence as low because of high heterogeneity and limited blinding. The best type and amount of exercise remain unclear.

Is mindfulness as effective as medication for anxiety?

In one randomized trial of 276 adults with anxiety disorders, an 8-week MBSR course was noninferior to escitalopram on a clinician-rated measure. The trial had no placebo arm and needs replication, and MBSR required an instructor-led course plus about 45 minutes of daily home practice.

Does slow breathing help with anxiety?

The evidence is mixed. Some reviews report small-to-medium benefits from regular breathwork, especially slow breathing, while a 2024 review found that brief single-session breathing exercises did not significantly reduce state anxiety compared with controls. It appears low-risk, but its effect size is uncertain.

Who should be screened for anxiety?

The US Preventive Services Task Force recommends screening for anxiety disorders in adults aged 64 and younger, including pregnant and postpartum people (Grade B). It found the evidence insufficient for adults 65 and older. Screening tools cannot diagnose an anxiety disorder on their own.

Sources

1. Any Anxiety Disorder (statistics) , National Institute of Mental Health
2. Social Anxiety Disorder (statistics) , National Institute of Mental Health
3. Generalized Anxiety Disorder (statistics) , National Institute of Mental Health
4. Anxiety Disorders Prevalence: 12-month Prevalence fact sheet , National Institute of Mental Health
5. How Many People Have Anxiety? Prevalence in the US and Worldwide , mhstats.org
6. Anxiety Disorders , National Alliance on Mental Illness
7. Cognitive behavioral therapy for anxiety and related disorders: A meta-analysis of randomized placebo-controlled trials , Depression and Anxiety (Wiley) (2018-01-01)
8. Cognitive-behavioral therapy for adult anxiety disorders in clinical practice: a meta-analysis of effectiveness studies , PubMed
9. Effect of cognitive-behavioral therapy for anxiety disorders on quality of life: a meta-analysis , PubMed
10. A meta-analysis on the efficacy of low-intensity cognitive behavioural therapy for generalised anxiety disorder , PubMed Central
11. Comparative efficacy of face-to-face and internet-based cognitive behavior therapy for generalized anxiety disorder: A meta-analysis of randomized controlled trial , Frontiers in Psychiatry (PubMed Central) (2022-01-01)
12. The effects of exercise on generalized anxiety symptoms in adults: a systematic review and meta-analysis , BMC Sports Science, Medicine and Rehabilitation (PubMed) (2026-01-01)
13. Effect of physical activity for reducing anxiety symptoms in older adults: a meta-analysis of randomized controlled trials , BMC Sports Science, Medicine and Rehabilitation (Springer) (2024-01-01)
14. The Effect of Exercise Training on Anxiety Symptoms in Children and Adolescents: A Systematic Review and Meta-Analysis of Randomized Trials , PubMed
15. Comparative effects of structured exercise protocols on depression and anxiety symptoms: a network meta-analysis , PubMed Central
16. Mindfulness-Based Stress Reduction vs Escitalopram for the Treatment of Adults With Anxiety Disorders: A Randomized Clinical Trial , JAMA Psychiatry (2022-11-09)
17. Mindfulness-Based Stress Reduction vs Escitalopram for the Treatment of Adults With Anxiety Disorders (abstract copy) , Ovid
18. Mindfulness-Based Stress Reduction vs Escitalopram for the Treatment of Adults With Anxiety Disorders (record with CGI-S results) , Semantic Scholar
19. Mindfulness Meditation vs Escitalopram for Treatment of Anxiety Disorders: Secondary Analysis of a Randomized Clinical Trial , JAMA Network Open (PubMed Central) (2024-01-01)
20. Treatments for Anxiety: Meditation and Escitalopram (NCT03522844) , ClinicalTrials.gov
21. Recommendation: Anxiety Disorders in Adults: Screening , US Preventive Services Task Force (2023-06-20)
22. Screening for Anxiety Disorders in Adults , American Family Physician (American Academy of Family Physicians)
23. Brief respiratory, embodiment, cognitive and mindfulness interventions to reduce state anxiety , Frontiers in Psychology (2024-01-01)
24. How Breath-Control Can Change Your Life: A Systematic Review on Psycho-Physiological Correlates of Slow Breathing , Frontiers in Human Neuroscience (2018-09-07)
25. Pranayama for mental disorders , Frontiers in Psychiatry (2025-01-01)
26. Review and meta-analysis suggests breathwork may be effective for improving stress and mental health , News-Medical (2023-01-13)

This article was drafted with the help of AI and reviewed by Jaroha editors before publication.