Da Xuan for All

Research on Taijiquan

Few traditional practices have been subjected to as much experimental scrutiny as taijiquan. Since the early 1990s, several thousand articles indexed in major medical databases have been devoted to it, and the rate of publication has continued to rise.

Here is a brief update…

The history of this research really began in Atlanta in the mid-1990s. Steven Wolf’s team, as part of the US FICSIT programme dedicated to the prevention of falls in older people, compared taijiquan training with balance board training and a control group receiving health education. The result, published in 1996, surprised everyone, including the authors: the slowest, least spectacular and least technical practice reduced the risk of multiple falls by almost half. This was not the most likely hypothesis, and it is precisely what gave this result its catalytic power. Over the following twenty-five years, fall prevention remained the area with the strongest evidence.

The evidence has accumulated to the point where it has become difficult to dispute. A 2023 meta-analysis combining twenty-four randomised trials found a reduction in the risk of falls of around a quarter among older people, whether they were in good health or already considered to be at high risk, with a corresponding decrease in the average number of falls per person. Another, published in early 2026 and based on twenty-one trials from international and Chinese databases, confirms improvements in balance, walking speed and confidence when moving about. One point keeps recurring in these reviews: the effect increases with the duration and frequency of the intervention. Twelve weeks produce some results, twenty-four weeks produce more, and stopping the programme causes the benefits to diminish. It is not a one-off treatment.

The most conclusive trial remains that by Fuzhong Li, published in 2018 in JAMA Internal Medicine. Six hundred and seventy adults aged seventy and over, all at high risk of falling, were allocated to either a taijiquan programme designed to prevent falls, a conventional multimodal exercise programme comprising strength training, aerobic exercise and balance training, or a stretching programme serving as an active control. Over a six-month period, the taijiquan group experienced significantly fewer falls than the stretching group, and significantly fewer than the conventional exercise group, despite the latter being well-designed. This is one of the rare instances where the practice does not merely perform as well as a serious alternative: it performs better.

This finding has institutional implications. The global guidelines on fall prevention published in 2022, as well as the Cochrane reviews on exercise and falls, include taijiquan amongst the interventions proven to be effective for people living at home. In other words, the practice has moved beyond the realm of so-called alternative medicine to become part of standard public health recommendations. Few traditional practices have crossed this threshold.

It remains to be understood why. The mechanisms identified are not particularly mysterious, which is rather reassuring. Taijiquan constantly involves shifting weight from one foot to the other, meaning that the practitioner spends much of their time on one foot—either partially or fully—in semi-bent positions that engage the quadriceps in a prolonged and eccentric manner. The slow pace, far from being a flaw, forces the practitioner to control every moment of their movement rather than letting themselves be carried along by momentum. Posturographic studies show an improvement in stability limits – that is, the range within which a person can shift their centre of mass without losing their balance – and better control within that range. Research into ankle and knee proprioception demonstrates greater precision in reproducing joint angles. Several research teams have observed a reduced reliance on visual information to maintain an upright stance, which is particularly important at night, in dim light, or when turning the head.

To this must be added a factor that protocols often underestimate: the fear of falling. This is a risk factor in its own right, independent of actual physical abilities, because it leads to reduced movement, and thus to a further loss of ability. However, trials measuring a person’s sense of self-efficacy in the face of the risk of falling consistently find significant improvements. A person who has spent three months slowly exploring the limits of their balance, knowing that they will not fall during the exercise, regains a confidence that no isolated muscle-strengthening exercise can provide.

The second major area is neurological, and is dominated by Parkinson’s disease. In 2012, Fuzhong Li published a trial in the New England Journal of Medicine involving one hundred and ninety-five patients divided into groups practising taijiquan, muscle strengthening and stretching, over a period of twenty-four weeks. The taijiquan group achieved better results in terms of stability limits, stride length, functional strength and the Motor Reference Scale, with fewer falls, and some of the benefits persisted three months after the end of the programme. Publishing this type of result in this journal changed perceptions of the subject within the medical community.

Since then, the focus has shifted to the long term. A team from Shanghai followed patients with early-stage Parkinson’s disease for one year, comparing taijiquan, brisk walking and no exercise, using functional imaging, cytokine assays and metabolic analyses. The authors report improvements in gait and balance, associated with changes in brain network connectivity, a reduction in inflammatory markers and alterations in amino acid metabolism. The same cohort, followed for three and a half years and published in 2023 in the Journal of Neurology, Neurosurgery and Psychiatry, suggests slower symptom progression, fewer complications and a smaller increase in medication doses among practitioners. This is a cohort study rather than a randomised trial over this timeframe, which means it is not possible to draw conclusions about an effect on the disease itself; however, the evidence is sufficiently consistent to justify the confirmatory trials currently underway.

Recent meta-analyses reinforce the picture regarding functional parameters: improved balance as measured on the Berg Scale, reduced time taken in the stand-up and walk test, and improved walking speed. A three-arm Chinese trial conducted between 2021 and 2023 also shows that both taijiquan and a combination of strength training and stretching improve motor function and quality of life at intermediate stages, which calls for a measure of caution: it is not the only effective approach, but it is an effective one, particularly suited to people who are unlikely to visit a gym.

Outside of Parkinson’s disease, the neurological evidence is more limited. In patients who have suffered a stroke, several trials and meta-analyses have found improvements in standing balance and a reduction in falls during the chronic phase, including with seated versions adapted for severe hemiplegia. In multiple sclerosis, a few small-scale trials show effects on balance and fatigue, but the sample sizes remain too small to be conclusive. In diabetic peripheral neuropathies, the results are preliminary and contradictory.

The third area, arguably the most significant in terms of the number of people affected, is that of chronic musculoskeletal pain.

Knee osteoarthritis is the best-documented case. In 2016, Chenchen Wang published a trial in the Annals of Internal Medicine comparing twelve weeks of taijiquan with twelve weeks of standard physiotherapy in 204 patients. Both approaches produced comparable improvements in pain and function, with taijiquan proving no less effective, whilst showing an advantage over standard physiotherapy in terms of depression scores and quality of life. This finding carried significant weight: the American College of Rheumatology’s guidelines, published in 2019, list taijiquan amongst the strongly recommended interventions for osteoarthritis of the knee and hip – an unusual position for a practice with martial arts origins. Subsequent meta-analyses, including one published in 2025 on knee pain, confirm a genuine effect on pain and function, with excellent safety.

The plausible mechanisms are, once again, fairly straightforward: strengthening of the quadriceps and hip stabilisers, improved joint proprioception, moderate and distributed joint loading rather than high-impact loading, better distribution of stresses through alignment, and reduced fear of movement. The height of the positions is a crucial factor here, which I shall return to, as a position that is too low in a patient with osteoarthritis of the knee produces exactly the opposite of the desired effect.

Fibromyalgia is the other notable case. In 2010, an initial trial by Chenchen Wang published in the New England Journal of Medicine, involving sixty-six female patients over twelve weeks, showed substantial improvements in pain, sleep, mood and quality of life. In 2018, the same team published a much more robust trial in the British Medical Journal: 226 patients, a direct comparison with supervised aerobic exercise – which is the standard non-pharmacological treatment – and follow-up for up to 52 weeks. Taijiquan performs at least as well, and on several measures better, with an effect that increases with duration of exposure: twenty-four weeks are more effective than twelve. The European guidelines on fibromyalgia also mention this practice amongst the possible non-pharmacological options.

Chronic low back pain also shows positive results. A 2025 meta-analysis of eight randomised trials involving 729 adults found effects on pain and disability, although the certainty of the evidence was limited by sample size and methodological quality. The 2017 US guidelines on chronic low back pain nevertheless include tai chi in the list of non-pharmacological approaches to be offered as first-line treatment, alongside yoga, exercise, cognitive therapy and acupuncture. This recognition is based as much on safety and acceptability as on the magnitude of the effect.
For rheumatoid arthritis, spondylitis or chronic tendinopathies, data do exist but are too scattered to draw a conclusion. It can be said that no warning signs are apparent and that patients generally experience functional improvement, though it is not possible to go into further detail.

The issue of bone health warrants separate mention, as it is often misrepresented. Several meta-analyses in postmenopausal women have found a favourable, albeit weak, effect on lumbar and femoral bone mineral density, with low-quality evidence. It would be disingenuous to present tai chi as a treatment for osteoporosis. However, the relevant issue is not bone density but fracture, and a fracture almost always involves a fall. From this perspective, the benefit of Taijiquan for people with osteoporosis lies primarily in the prevention of falls, which is a far stronger argument than a variation of a few per cent in a bone density index.

In terms of cardiovascular health, the accumulation of data now allows for quantified estimates. A comprehensive review published in early 2026, combining thirty-nine randomised trials and twenty-one systematic reviews, puts the average reduction in systolic blood pressure at around six millimetres of mercury and the reduction in diastolic blood pressure at around three. These figures are of the same order as those obtained with other forms of moderate exercise, and they are clinically significant at a population level: a sustained reduction of this magnitude results in a measurable reduction in the risk of stroke. A meta-analysis published in 2026 on women aged 45 and over found, in addition to the reduction in blood pressure, reductions in LDL cholesterol, triglycerides and total cholesterol, with no clear effect on HDL or fasting blood glucose levels.

In heart failure, the best-known trial remains that of Gloria Yeh, which demonstrated, in around a hundred patients with reduced ejection fraction, an improvement in quality of life, mood and confidence during exertion, without any improvement in the distance covered in six minutes. This result is interesting precisely because of this disconnect: what patients experience and what performance measures capture do not always coincide, and it is not clear that performance measures are the most relevant criterion for them.

In post-myocardial infarction rehabilitation, several studies highlight, above all, the importance of acceptability: patients who refuse conventional cardiac rehabilitation accept a Tai Chi programme and subsequently increase their overall physical activity. This is a general point worth emphasising. In many populations, the limiting factor is not the theoretical efficacy of an exercise but the likelihood that someone will still be doing it in six months’ time. The adherence rates reported in taijiquan trials are often higher than those of conventional programmes, and this forms part of its actual efficacy.

Breathing is a less well-known yet robust area. In chronic obstructive pulmonary disease, several meta-analyses have found an improvement in the distance covered in the six-minute walk test of between thirty and fifty metres – which in several trials exceeds the threshold for clinical significance – as well as an improvement in respiratory quality of life scores and a reduction in dyspnoea. A comparative trial conducted in China and published in Chest even suggested equivalence with conventional respiratory rehabilitation, with greater sustained benefits twelve weeks after the end of the programme – a finding likely explained by the fact that participants can continue practising what they have learnt independently at home. The mechanisms involved relate to the coordination of movement and breathing, diaphragmatic breathing, prolonging exhalation and reducing anxiety associated with shortness of breath.

Since 2020, several teams have applied these protocols to rehabilitation following prolonged respiratory infection, with encouraging results regarding fatigue and functional capacity; however, these findings are still too recent and too heterogeneous to draw any firm conclusions.

The cognitive domain is where enthusiasm most often outstrips the evidence, which does not mean there is nothing to be gained. Meta-analyses focusing on mild cognitive impairment find improvements in overall scores and, more interestingly, in executive functions – that is, planning, inhibition and mental flexibility. A review published in 2025 in Frontiers in Physiology, limited to taijiquan interventions alone in adults aged 60 and over with mild cognitive impairment, concluded that there was a benefit in certain areas whilst calling for higher-quality research. Imaging studies, notably those conducted in Shanghai on long-term practitioners, describe differences in volume or cortical thickness in regions associated with sensorimotor integration and interoception, but these are most often cross-sectional comparisons between practitioners and non-practitioners, which do not allow us to distinguish the effect of the practice itself from the fact that certain individuals engage in it.

The psychological field is more favourable. A meta-analysis from 2024 on depression in older adults found a substantial improvement in scores on the Geriatric Depression Scale, with a significantly greater effect for six-month programmes than for shorter programmes. With regard to anxiety and perceived stress, the effects are more modest but consistent. As for overall quality of life, improvement is one of the most consistently replicated findings across the entire literature.

Sleep warrants further discussion. Michael Irwin and his team have conducted a series of remarkable studies on this topic. As early as 2008, they demonstrated an improvement in sleep quality amongst older people. In 2017, in the Journal of Clinical Oncology, they compared taijiquan with cognitive behavioural therapy for insomnia in breast cancer survivors suffering from severe insomnia: taijiquan was not inferior to the standard of care, and the effect was sustained for fifteen months. A team from Hong Kong subsequently extended this finding to older adults with insomnia, with a two-year follow-up. This is one of the few areas where it can be said that the practice compares favourably with a validated psychotherapeutic treatment, and not merely with no treatment at all.

In oncology, a systematic review published in 2025—which searched twenty-five databases and synthesised several previous meta-analyses—concluded that there was an improvement in patients’ overall quality of life, physical function, fatigue and sleep, as well as mental health and anxiety, though the findings regarding depression were less clear-cut. We must be perfectly clear about the scope of these findings: they relate to patients’ experiences, function and symptoms, in the context of treatment. There is no evidence to suggest an effect on survival or on the disease itself, and any presentation suggesting otherwise is both false and dangerous.

If we now attempt to bring together the mechanisms common to all these areas, we arrive at a fairly coherent picture.

The first level is biomechanical. Continuous weight-shifting, alternating weight-bearing, prolonged work in a semi-flexed position, eccentric control, coordinated rotations of the pelvis and trunk, and maintained joint range of motion without forced passive stretching. This results in moderate strengthening of the lower limbs, improved hip and ankle stability, and preserved spinal mobility, all with low joint stress and without impact. For a seventy-year-old, or for a damaged knee, this is a compromise that is hard to match.

The second level is sensorimotor. The slow pace allows for dense and continuous proprioceptive feedback, whereas rapid movement is ballistic and eludes correction. Alignment requires an internal representation of the body that becomes more refined with repetition. Studies of posturography, gait variability and reaction times all point to better integration between proprioceptive, vestibular and visual information, with less reliance on vision. This is precisely what deteriorates with age, and it is precisely what causes falls.

The third level is autonomic. Several studies report an increase in heart rate variability and indices of parasympathetic tone during and after practice. Slow breathing, often approaching six cycles per minute in experienced practitioners, falls within the baroreflex resonance zone, which is sufficient to explain some of the effects on blood pressure and anxiety without the need to invoke any other factors. It is interesting to note that the tradition emphasises relaxation, stillness and the continuity of the breath long before any devices were available to measure them.

The fourth level is attentional and psychological. The practice requires sustained but relaxed attention, directed towards bodily sensations. Structurally, this is similar to what the literature refers to as mindfulness, with the difference that it is practised whilst moving, which makes it accessible to people who would not be able to sit still. The effects on rumination, perceived stress, self-efficacy and body image likely stem from this.

The fifth level is social, and it is systematically underestimated because it troubles methodologists. The practice takes place in a group, at a fixed time, in a regular venue, under the supervision of a teacher, often for years on end. For someone living in isolation, it is sometimes the only group activity of the week. Trials that use passive control groups therefore measure this effect to some extent, and some commentators view this as a bias. One might just as well say that it forms part of the practice’s actual effectiveness in real life, and that it would be absurd to seek to exclude it.

The question of dosage is better understood than is commonly believed. Effective protocols almost always involve between two and three weekly sessions of around one hour, for at least twelve weeks, with more pronounced effects at twenty-four weeks and beyond. Short daily home exercises as a complement improve the results. The most consistent finding – and the most troublesome for marketing claims – is that the effect wears off when the practice ceases. Long-term follow-ups show a gradual decline in the benefits to balance and function among those who have stopped practising. There is no permanent gain. It is a lifestyle habit, not a cure.

This leads to the question that everyone eventually asks: is taijiquan superior to other forms of exercise?

The honest answer is YES.

In other words, the uniqueness of taijiquan does not lie in an ‘active ingredient’ found nowhere else. It lies in a rare combination: balance, gentle strengthening, coordination, mindfulness, breathing, low-impact nature, no equipment required, the ability to practise at any age and in a small space, and a capacity for sustained engagement that few activities achieve. For an ageing or pain-afflicted population, this combination is more valuable than a single, slightly superior benefit.

A word on safety, which is a strength that is rarely emphasised. Adverse effects reported in trials are rare and mild: muscle soreness, temporary knee or lower back pain, mainly at the start of the programme and primarily linked to postures that are too low. Withdrawals due to adverse effects are exceptional. Compared with most pharmacological interventions or certain intensive exercise programmes, the risk-benefit ratio is excellent. All this stems from a form of exercise that is poorly taught by people with limited competence.

Nevertheless, a few precautions are worth reiterating. A posture that is too low is the primary cause of knee pain in beginners, and the height must be adjusted to suit the individual, not for the sake of the form’s aesthetics. Rotating the body whilst keeping the knee fixed and the foot locked to the floor produces a shearing force that no joint can tolerate. People suffering from positional vertigo, severe neuropathy or significant instability should start with a support available. And practising is not a substitute for ongoing treatment: none of the studies cited tested stopping medication; all were conducted alongside standard care.

And we can tell them the essential point, which is often overlooked in protocols: that the reason people carry on for thirty years is almost never the same as the reason they started. People come for their back, they stay for something else. Quality-of-life scales capture a distant echo of that ‘something’, but they cannot put a name to it.

Health-focused taijiquan and martial taijiquan are often presented as two options between which one must choose: one gentle and sensible, the other demanding and somewhat superfluous. This is a misunderstanding, and it comes at a high cost to those who practise solely to keep fit, because the martial dimension does not merely add a purpose to the practice: it ensures its correctness.

The first reason is structural. A posture without application becomes distorted. No one notices this until there is contact, because there is nothing to penalise an inaccurate posture. As soon as a hand is placed on you, the alignment is automatically verified; the support is revealed to be correct or incorrect; and ‘rooting’ ceases to be a metaphor. Yet all the documented physical benefits—for the knees, the back and balance—depend precisely on this quality of alignment. Martial arts training is not an ‘extra’ for the soul; it is the quality control of what we do the rest of the time.

The second reason concerns balance itself. When practising alone, we work on anticipated balance: we know where we’re going, we choose our pace, we prepare our weight transfer. In real life, we fall for the opposite reason – because a pavement, a hand or a dog appears without warning. Contact introduces precisely that: an external disturbance that we do not choose, to which we must respond whilst maintaining our structure. It is a level of balance that solitary practice never achieves, and it is the one that counts when it really matters.

The third reason is one of attention. Memorising a sequence already demands attention and motor memory. Reading a partner’s intention, sensing where they are full and where they are empty, deciding without deliberation, whilst remaining relaxed, adds a whole new dimension of challenge. We move from a repetitive task to an open-ended one, which is far more demanding and far more alive.

The fourth reason is emotional, and it is perhaps the most important. Relaxing on your own, in a quiet room, with your eyes half-closed, is not difficult. Remaining relaxed when someone pushes you, takes you by surprise, puts you in a difficult position, when fear rises, when pride makes you tense up – that is the true work of relaxation. Within seconds, contact reveals all the tension you thought you had released. Learning to release them under pressure, rather than at rest, is what carries over into the ordinary situations of life, where no one gives you time to sit down and catch your breath.

There is an irony in all this that I find rather pleasing. Tradition has never separated health from the rest. The ‘Song of the Thirteen Postures’ concludes by asking what, ultimately, is the point of all this work of intention, and answers it with longevity – a spring that never grows old.

This was not a therapeutic promise in the modern sense; it was the assertion that a proper discipline, as a bonus, takes care of life without making it its primary aim. Thirty years of Western methodology have verified, with their instruments and their limitations, part of what this statement foretold. They have not verified the rest, and it is not certain that they ever will.

Which, from a practical point of view, is ultimately of no importance. We have never needed a randomised trial to carry out a weight transfer correctly. But when a doctor asks what we base this on, it is nice to be able to give an answer other than ‘tradition’, and to quote figures.

And it is always one more reason to give when you are asked: ‘But why are you doing that?’