Tai Chi

What it is

Tai chi is a traditional Chinese martial art [1]. It is also described as a mind-body exercise [2]. Trials have used several styles and forms, such as Yang style, Sun style and the 24-form simplified routine [3] [4] [2]. Most trials ran as group classes, but one trial delivered it as online video at home [5].

How it might work

A proposed pathway for the blood pressure effect is lower angiotensin II and higher nitric oxide. In a 99-person trial the blood pressure drop went together with both changes, but this has not been confirmed [1]. The mindful elements of tai chi are another proposed reason for its mood effects, though a review notes they have not yet been measured [6].

What the research says

Falls: a 2023 meta-analysis of 24 randomized trials found that tai chi lowered the risk of falls in older adults (rate ratio 0.76, 95% CI 0.71 to 0.82). It also improved balance tests, including the timed up-and-go test [7]. In a 670-person trial of older adults at high risk of falling, a tailored tai ji quan program had an incidence rate ratio of 0.42 for falls compared with stretching [8]. A 503-person trial in older adults aged 70 and over found no difference in falls [4].

Knee osteoarthritis pain: three meta-analyses, covering 407 to 986 people, found less pain and better physical function with tai chi [9] [10] [11]. One of them found a pooled pain effect of -0.69 standard deviations across 986 people [9]. In a 204-person head-to-head trial, tai chi and physical therapy gave similar improvements at 12 weeks, and the benefit held to 52 weeks [12]. A 2026 trial of online tai chi found a larger drop in knee pain than an information-only control: a mean difference of -1.4 points on a 0 to 10 scale [5].

Fibromyalgia: a meta-analysis of 6 trials (657 people) found that tai chi reduced fibromyalgia impact scores more than standard care (standardized mean difference -0.61), and reduced pain [13]. In a 226-person trial, tai chi improved the revised impact score by 5.5 points more than aerobic exercise over 24 weeks, and by 16.2 points more in the twice-weekly group [3].

Blood pressure: a meta-analysis of 31 controlled trials (3,223 people, mostly with prehypertension) found tai chi lowered systolic pressure by 11.3 mmHg and diastolic by 4.8 mmHg compared with control [14]. The authors report that English-language trials showed about half the effect of Chinese-language trials, so the pooled figure may be overstated [14]. In a 342-person trial of adults with prehypertension, tai chi lowered systolic pressure 2.40 mmHg more than aerobic exercise after 12 months [15]. A 2024 meta-analysis of five small trials found a significant systolic effect but no significant diastolic effect [16].

Sleep quality: two meta-analyses of 20 trials each found better sleep quality with tai chi, with 1,703 patients in one [17]. The other pooled traditional Chinese exercise, which includes tai chi and qigong, and found a Pittsburgh Sleep Quality Index improvement of 2.24 points [18]. In a 90-person trial in breast cancer survivors with insomnia, tai chi chih was not worse than cognitive behavioral therapy for insomnia: 46.7% reached a treatment response at 15 months, compared with 43.7% for therapy [19].

Anxiety and depression: a 2023 meta-analysis of 23 studies (4,370 people) compared tai chi with other non-mindful exercise and found small improvements in anxiety, depression and general mental health [6]. The depression effect was small (d = 0.20) and the authors call the evidence tentative [6]. A separate meta-analysis of 10 trials found lower depression scores with tai chi than with conventional therapy (standardized mean difference -0.58) [20].

Thinking and memory: a meta-analysis of 33 trials (1,808 people) found better Montreal Cognitive Assessment (MoCA) scores with tai chi, with a mean difference of 3.23 points [21]. In people with dementia, the effect was smaller (standardized mean difference 0.27 across 7 trials), with no effect on depression scores or the timed up-and-go test [22]. In a 328-person trial of older adults with type 2 diabetes and mild cognitive impairment, tai chi chuan scored 0.84 MoCA points higher than fitness walking at 36 weeks [2]. In 318 older adults with mild cognitive impairment or memory concerns, a cognitively enhanced tai ji quan program improved MoCA by 1.5 points more than standard tai ji quan and by 2.8 points more than stretching [23].

Doses used in studies

These are the doses used in trials, not recommendations. Falls trials used 60-minute classes twice a week, for 48 weeks in people aged 70 and over [4], and for 24 weeks in people at high risk of falling [8]. In knee osteoarthritis, one trial used twice-weekly sessions for 12 weeks [12]. One review found that programs longer than 16 weeks, three times a week, worked best for pain and physical function [11]. Cognition trials used 60-minute sessions three times a week for 24 weeks [2], or one-hour sessions twice a week for 24 weeks [23]. A blood pressure trial ran four 60-minute sessions a week for 12 months [15].

Safety and interactions

Common side effects: the trials that reported them found no serious adverse events with tai chi [12] [3]. Few trials recorded adverse events in a structured way. A review of tai chi and qigong in brain injury found that only one trial reported adverse events, with none in either group [24]. In a falls trial of people with some disability, more people in the tai chi group stopped attending in the first 24 weeks (difference 17.9 percentage points) [4].

Upper intake limits: none apply, and no toxic dose was reported. Medicine interactions: none of the studies found reported an interaction with medicines, and the review of adverse events data notes that safety data are still limited [20]. Pregnancy and breastfeeding, kidney or liver disease, and bleeding disorders were not addressed in the studies found for this page, so these studies do not establish safety for those groups [20].

Caveats and open questions

The falls result is mixed. The 503-person trial found no effect, and its authors suggest that low class attendance or attrition may have contributed [4]. The strongest falls trial reported that its lead author founded the company that licenses the program tested, and receives a licensing fee [8].

In the knee and fibromyalgia trials, participants knew which treatment they were getting, so self-reported pain and mood may be affected by expectations [12] [20]. Most knee and mood trials are small, and one review notes that large-sample studies are limited [11].

The blood pressure and sleep findings are also uncertain. The pooled blood pressure effect may be larger than the true effect because of differences between English- and Chinese-language trials [14]. The sleep reviews are inconsistent and one combines tai chi with qigong [18].