Deep Breathing for Anxiety: Techniques and What Trials Show

Deep breathing for anxiety has measurable but modest support from randomized trials. A 2023 meta-analysis in Scientific Reports pooled 20 trials and found breathwork was associated with lower self-reported anxiety than control conditions, with a small-to-medium effect size of g = -0.32^[1]. The same analysis found similar effects for stress (g = -0.35 across 12 trials and 785 adults) and depressive symptoms (g = -0.40 across 18 trials)^[1].

Those results apply to symptoms of anxiety in mostly non-clinical groups, not to anxiety disorders as diagnosed conditions. The stress results were significant in nonclinical samples but not in mental or physical health populations, and the mental health subgroup contained only one study^[1].

A 2021 meta-analysis in Clinical Psychology Review took a wider view: 40 randomized trials of interventions with a respiratory component, measured against trait anxiety^[2]. It reported a medium-to-large average effect (g = -0.678) and concluded that breathing interventions have clinical utility either on their own or as an add-on to other treatment^[2].

The size of the benefit depended on the comparison. Effects were stronger against inactive controls such as waitlists than against active treatments, and heterogeneity across studies was high^[2]. The 2023 authors put it plainly, warning against a "miscalibration between hype and evidence"^[1].

Breathing changes are part of how anxiety shows up in the body. The 2021 meta-analysis describes respiratory abnormalities as "a hallmark of anxiety symptomatology," and notes that standard anxiety treatments often do not address breathing directly^[2].

The link is clearest in panic attacks. The National Institute of Mental Health lists difficulty breathing among the physical symptoms of a panic attack, along with a pounding or racing heart, chest pain, dizziness and tingly or numb hands^[8].

Slow breathing works in the opposite direction. Slow, deep breathing increases parasympathetic activity, the branch of the nervous system that calms the body, according to the authors of a 2022 physiology study^[5]. In one controlled breathwork trial, daily practice reduced respiratory rate more than mindfulness meditation did^[3].

That physiological shift is the theory behind the techniques. Whether it translates into fewer anxiety symptoms over weeks or months is a separate question, and the trial evidence on that is weaker than the physiology.

Most studied techniques vary three things: the pace of breathing, the length of the exhale relative to the inhale, and whether the breath is held.

Diaphragmatic (belly) breathing

Slow, deep breaths that move the abdomen. A 2019 review of 3 studies with 880 participants found preliminary evidence that diaphragmatic breathing may reduce stress, with changes in self-rated mental health and in physical measures such as cortisol and blood pressure^[7].

Box breathing

Inhalation, breath hold and exhalation of equal length^[3]. Direct trial data on box breathing are limited; the main example is the one-month Stanford comparison described below.

Cyclic sighing

A pattern that emphasizes prolonged exhalations^[3]. It produced the largest mood gains of the techniques in the 2023 Stanford trial^[3].

The 4-7-8 technique

Inhale through the nose for a count of 4, hold for 7 seconds, then exhale through the mouth for a count of 8^[5]. The method is based on yogic pranayama breathing and is promoted for anxiety and sleep^[5].

Slow-paced breathing and HRV biofeedback

Heart rate variability (HRV) biofeedback uses a device that displays heart rate data while the person uses breathing techniques to change the heart rate pattern. A 2017 review of 24 studies with 484 participants found it helped reduce self-reported stress and anxiety^[7]. In the 2023 meta-analysis, slow-paced breathwork reduced stress while fast-paced breathwork did not^[1].

The best-known recent trial came from Stanford in 2023^[3]. Participants did one of three 5-minute daily breathing exercises (cyclic sighing, box breathing or cyclic hyperventilation with breath retention) or 5 minutes of mindfulness meditation, every day for a month^[3].

All groups showed significant reductions in state anxiety and negative mood^[3]. Breathwork, especially cyclic sighing, produced a greater improvement in mood and a larger drop in respiratory rate than meditation^[3].

On anxiety specifically, breathwork did not beat meditation. The average daily drop in state anxiety was -3.85 points for cyclic sighing, -3.75 for box breathing and -3.95 for mindfulness meditation^[3].

The authors listed the limits themselves: the study was registered retrospectively, adherence was self-reported in a fully remote design, there was no follow-up beyond 4 weeks, and each group was small^[3].

The meta-analyses carry similar caveats. Most trials in the 2023 review were at moderate risk of bias, and there were too few follow-up measurements to analyze lasting effects^[1]. The 2021 review found significant unexplained variation between studies, which it said may reflect differences in intervention design, population and control groups^[2].

No trial shows that breathing exercises cure an anxiety disorder. The evidence supports symptom reduction, mostly measured over weeks, in people who are often not in treatment.

For anxiety disorders, the comparison that matters is cognitive behavioral therapy (CBT). The National Institute of Mental Health describes CBT as the well-studied "gold standard" psychotherapy for panic disorder, and notes that exposure therapy may be used along with relaxation exercises and breathing training^[8].

Direct comparisons of relaxation-based approaches and CBT disagree. A 2018 review of 16 studies with 856 people with generalized anxiety, social anxiety or panic disorder found relaxation therapy more effective than CBT for reducing anxiety^[7]. Another 2018 review of 50 studies and 2,801 people found relaxation less effective than CBT for PTSD and OCD and no difference for other anxiety disorders, with most studies at high risk of bias^[7].

NCCIH summarizes the evidence by saying CBT may be more helpful than relaxation techniques for at least some anxiety disorders, and that relaxation may be less effective after one year for panic disorder^[10]. Details on CBT and other options are in our guides to therapy for anxiety and anxiety medication.

Breathing retraining has long been part of CBT for panic disorder. A component network meta-analysis of 72 trials with 4,064 participants tested which pieces of CBT for panic disorder carry the benefit^[4].

Interoceptive exposure, which deliberately brings on feared body sensations, and face-to-face delivery were associated with better results^[4]. Breathing retraining appeared to improve acceptability but had small effects on efficacy, and muscle relaxation was associated with significantly lower efficacy^[4].

The authors raised but could not settle one question: whether breathing retraining acts as a counterproductive "safety behaviour" during exposure. Their data were too sparse to detect that kind of interaction^[4]. They also stressed that every component, breathing retraining included, outperformed a waitlist^[4].

In trials limited to psychiatric samples, the 2021 meta-analysis still found a moderate effect of respiratory interventions on anxiety (g = -0.513 across 13 trials)^[2]. How a panic attack differs from the looser term anxiety attack is covered in panic attack vs. anxiety attack.

The most recent national figure for deep breathing specifically is dated. In the National Health Interview Survey, deep-breathing exercises were the second most commonly used complementary health approach among U.S. adults in 2002 (11.6%), 2007 (12.7%) and 2012 (10.9%)^[6].

The 2012 figure is not directly comparable to earlier years: that survey collected deep breathing as part of other approaches, such as meditation, yoga and guided imagery, rather than as a stand-alone question^[6].

Newer data cover related techniques. In 2022, 6.4% of U.S. adults used guided imagery or progressive muscle relaxation, up from 3.8% in 2002^[7].

The potential audience is large. An estimated 19.1% of U.S. adults had an anxiety disorder in the past year, based on National Comorbidity Survey Replication data from 2001 to 2003^[9]. Current figures by age, sex and state are on our anxiety statistics and generalized anxiety disorder pages.

Relaxation techniques, including breathing exercises, are generally considered safe for healthy people, and most studies report no negative side effects^[7].

Occasionally, people report increased anxiety, intrusive thoughts or fear of losing control. There are also rare reports of relaxation techniques causing or worsening symptoms in people with epilepsy, certain psychiatric conditions, or a history of abuse or trauma^[7].

Pace may matter for benefit as well as comfort: in the 2023 meta-analysis, slow-paced breathwork reduced stress and fast-paced breathwork did not^[1]. Body-focused approaches that go beyond breathing are covered in our article on somatic therapy.