Moxibustion heats an acupuncture point by burning moxa, the dried and processed leaf of mugwort,
on the skin or just above it. A 2013 review of its history and mechanism by
Hongyong Deng and Xueyong Shen traces it to the silk manuscripts found in the Mawangdui tomb, sealed
around 168 BC, which already prescribe it, and to the Huangdi Neijing, which discusses it
at length. Deng and Shen name the plant as Artemisia argyi; the US Council of Colleges of
Acupuncture and Oriental Medicine’s Clean Needle Technique Manual (seventh
edition, revised 2017) names Artemisia vulgaris as the most commonly used source, and the Italian breech trial below used sticks of A. vulgaris. One use has a
Cochrane review of its own: burning moxa over BL67 (Zhiyin),
at the tip of the little toe, to turn a baby lying bottom-first.
Forms of moxibustion
Deng and Shen divide moxibustion by whether the burning moxa touches the skin. Direct moxibustion places a cone on the skin and lights it. Indirect moxibustion keeps the moxa at a distance, with air or a slice of insulating material between: garlic, ginger, salt or a cake of aconite. Moxa comes as loose wool, as cones of different sizes and, most often according to the Clean Needle Technique Manual, as a rolled stick, in a traditional form and a smokeless one.
| Form | How it is done |
|---|---|
| Direct, non-scarring | A cone burns on the point and is removed when about two thirds has burnt or the patient feels burning; the skin should flush but not blister (Clean Needle Technique Manual) |
| Direct, scarring | The cone is left until it burns out or a blister forms; the manual asks for written consent first. Deng and Shen record that ancient physicians favoured it |
| Indirect, on an insulating layer | A cone burns on a slice of ginger or garlic, a bed of salt or an aconite cake, and is removed with the insulator when about two thirds has burnt |
| Moxa stick | A lit roll is held above the point; in one breech trial, 1.5 to 3 cm from the skin for 10 minutes on each foot (Higashihara and Horiuchi 2021) |
| Warm needle | A lit piece of moxa stick sits on the handle of an inserted needle (Yang and colleagues 2024) |
| Moxa box | Lit sticks burn inside a box laid over the skin, in one protocol over the navel with a cotton cloth between box and skin (Zhu and colleagues 2025) |
Deng and Shen report measurements, mostly in mice, of how hot each form gets: one direct cone raised the skin surface to about 130°C, while indirect moxibustion peaked around 65°C on the skin, and a ginger slice buffered the rise and fall in temperature.
Breech presentation: the Cochrane review
The most recent Cochrane review, updated in 2023 by Meaghan Coyle, Caroline Smith and Brian Peat, collected 13 randomised or quasi-randomised trials with 2,181 women whose babies were in breech position, searched to November 2021. Its main comparison set moxibustion added to usual care against usual care alone, and the table below reports it; the review also compared moxibustion with acupuncture and with postural techniques. One outcome was the need for external cephalic version (ECV), turning the baby by hand from outside the abdomen. The review graded each result for certainty using the GRADE system.
| Outcome | Trials (women) | Risk ratio (95% CI) | Certainty |
|---|---|---|---|
| Compared with usual care alone | |||
| Baby not head-down at birth | 7 (1,152) | 0.87 (0.78 to 0.99) | Moderate |
| Needed ECV | 4 (692) | 0.62 (0.32 to 1.21) | Low |
| Caesarean section | 6 (1,030) | 0.94 (0.83 to 1.05) | Moderate |
| Oxytocin used before or during labour | 1 (260) | 0.28 (0.13 to 0.60) | Moderate |
| Compared with sham moxibustion | |||
| Baby not head-down at birth | 1 (272) | 0.74 (0.58 to 0.95) | Moderate |
| Caesarean section | 1 (272) | 0.84 (0.68 to 1.04) | Moderate |
A risk ratio below 1 means the outcome was less common with moxibustion. The authors concluded that moxibustion added to usual care probably reduces the chance that the baby is not head-down at birth but probably makes little or no difference to the caesarean rate, and that adverse events were inadequately reported in most trials. The one trial whose adverse event data they could reanalyse recorded events, including nausea, an unpleasant smell, abdominal pain and contractions, in 27 of 65 women given moxibustion and none of 57 controls, a result they rated very low certainty.
The trials behind the review differ. An Italian trial by Francesco Cardini and colleagues, published in 2005, stopped at 123 women, under half its planned size, because many women stopped treatment, and found cephalic presentation at 35 weeks in 34% of the moxibustion group and 36% of controls. Its authors wrote that it supported neither the effectiveness nor the ineffectiveness of moxibustion. BL67 is also one of the points traditionally forbidden in pregnancy, a conflict Pregnancy and the forbidden points sets out, and AcuiQ’s breech presentation page lists the protocols individual studies prescribed, each with its citation.
Other reviews
Two reviews of other uses grade their own evidence. A 2022 overview by Shao Yin and colleagues gathered 10 systematic reviews of moxibustion for knee osteoarthritis, covering 57 trials and 5,149 participants. All 10 reviews had more than one critical flaw on the AMSTAR-2 appraisal tool, and of the 32 outcomes graded, 2 were high certainty, 3 moderate, 17 low and 10 very low. A 2024 review of warm needle acupuncture for allergic rhinitis by Shasha Yang and colleagues pooled 23 trials with 2,230 participants, none of which blinded patients or practitioners.
Burns and other adverse events
A 2010 systematic review of adverse events by Ji-Eun Park, Edzard Ernst and colleagues found reports in 4 randomised trials, 1 other controlled trial, 2 uncontrolled studies, 1 prospective study and 13 case reports. The events reported most were allergic reactions, burns and infections; trials also recorded redness, blistering, itching, discomfort from the smoke, tiredness, stomach upsets and headaches. The authors could not estimate how often any of these occur. The Clean Needle Technique Manual advises against direct moxibustion on the face or within the hairline, warns that people with diabetes or reduced sensation from nerve damage burn more easily, and says the practitioner should stay in the room throughout. Burns from moxibustion and cupping covers burns in full.
The smoke
Moxa smoke carries fine particles. A 2016 study by Chung-Yen Lu and colleagues measured the air in the moxibustion room of a Taipei hospital clinic. During moxibustion, fine particles (PM2.5) averaged 296 micrograms per cubic metre, against 26 before it began; Taiwan’s indoor air standard is 35, averaged over 8 hours. An extraction hood at the source removed 96% of the fine particles; a HEPA air cleaner in the room removed 71% and left levels above the standard; and burning aged moxa wool cut formaldehyde and volatile organic compounds by 75% and 83% but not the particles. The Clean Needle Technique Manual compares the risk to that of any other smoke, rising with total exposure over a working day, and asks for ventilation or HEPA filtering and for alternatives for patients with significant asthma.
Smokeless sticks are one response. In a 2021 Japanese pilot study of 60 women with breech babies, allocated to three groups without randomisation, 12 of 20 babies in the smokeless group, 9 of 20 in the smoke group and 5 of 20 controls were head-down at the end of treatment; only the smokeless group differed significantly from control, and there was no difference at birth.
What the sources cannot tell you
Only one trial in the breech review used sham moxibustion, and the Cochrane authors could judge harms from only one trial’s data. The knee osteoarthritis and rhinitis reviews pool trials whose quality their own authors rate low. The smoke measurements come from a single clinic room. A protocol entry on AcuiQ names moxibustion as its modality where the study used it, so a moxa stick at BL67 does not read as a needle protocol; Reading a protocol explains each field.