The circular purple marks that appeared on Olympic swimmers’ shoulders a decade ago did more for cupping’s visibility than two thousand years of practice had managed. They also created a lasting confusion, because the procedure those athletes received and the procedure most of the research tested are not the same thing.
Cupping comes in two forms. Dry cupping applies suction to skin that stays intact. Wet cupping deliberately punctures or scratches the skin first, so that blood is drawn into the cup. It is a bloodletting procedure.
In the largest review of the trial literature, covering 135 randomized trials published between 1992 and 2010, 78 of them, or 57.78 percent, used wet cupping as the main intervention. Western consumers at a sports clinic or a wellness studio are almost always offered the dry version.
So when a practitioner, a newspaper or a product page tells you that research supports cupping, the first question to ask is which cupping. The documented benefits and the documented harms both sit disproportionately on the side of the procedure most American clients never receive.
Two procedures, one word
A dry cupping session is straightforward to describe. A cup of glass, bamboo, silicone or earthenware is placed on the skin and the air inside it is withdrawn, either by a pump or by briefly burning something inside the cup before placing it, which is where the term fire cupping comes from. The cup may sit still for several minutes, be slid across oiled skin, or be applied and removed quickly in sequence. The suction pulls skin and surface tissue upward and ruptures small vessels, which is what produces the round bruise. The marks are not a sign of a problem and they are not a sign of a benefit. They are the predictable result of applying negative pressure to skin.
Wet cupping adds a step in the middle. The skin is punctured or lightly scratched, usually with a lancet or a fine blade, and suction then draws blood out through those openings into the cup. The federal National Center for Complementary and Integrative Health describes the difference in one line: in wet cupping “the skin is pierced, and blood flows into the cup.”
Everything that follows depends on holding those two apart. They share a name, a set of equipment and a tradition. They do not share a risk profile, and they are not equally represented in the research.
The harm record belongs mostly to wet cupping
Documented harm comes first here, because that is where the evidence is most specific and because it is the part consumer coverage routinely leaves until the end.
In July 2011, a team at Chungnam National University School of Medicine in South Korea published a prospective study of 206 adult men who presented with iron deficiency anemia. Each had a history taken and a physical examination, most had an endoscopy, and colonoscopy followed if no cause was found or a stool test came back positive. The point of the study was to work out what causes anemia in adult men, a group in whom it is less common and more likely to signal something serious.
Two histories stood out as probable causes. Previous stomach surgery accounted for 24 patients, 11.7 percent. Bloodletting cupping accounted for 11 patients, 5.3 percent.
Look at what else was in that table. Sixty-eight men, a third of the group, had an upper gastrointestinal disorder, including 34 peptic ulcers and 16 gastric cancers. Colonoscopy found clinically relevant lesions in 42. Among men aged 50 and over, 22 malignant lesions turned up, 16.8 percent of that age group. Wet cupping was sitting in the same list as stomach cancer as an explanation for why a man was short of iron.
The authors’ closing recommendation was that culture-specific procedures need to be considered as possible causes of iron deficiency anemia. There is a second-order risk buried in that sentence. An anemia confidently attributed to cupping is an anemia whose cancer work-up may stop early.
The National Center for Complementary and Integrative Health, the federal agency inside the National Institutes of Health that studies these practices, keeps a cupping page whose content is dated November 2018. It names burns, infections, scarring and skin discoloration that persists as side effects, and notes that cupping can worsen eczema or psoriasis, and records two rare severe outcomes: bleeding inside the skull after cupping on the scalp, and anemia from blood loss after repeated wet cupping. On equipment, the agency warns that gear used on more than one person without being sterilized in between can transmit hepatitis B and C.
A 2013 systematic review of published case reports, written by a team spanning Shanghai Municipal Hospital of Traditional Chinese Medicine and the University of Maryland School of Medicine, collected adverse events reported for acupuncture, moxibustion and cupping between 2000 and 2011. It found 117 reports describing 308 adverse events across 25 countries and regions. Ten of those events involved cupping. Infection was the leading complication across the whole family of practices, and in this period most infections were bacterial, arising from skin contact at treatment sites, with no hepatitis cases at all. The same authors had found hepatitis from reusable needles to be the main problem a decade earlier, which is a documented improvement worth noticing.
Ten published cupping cases in twelve years sounds reassuring. It should not be read as an incidence rate. It is a count of what got written up, in a field with no surveillance system, for a practice used by millions of people.
The most practical thing on the federal page is also the least quoted. Cupping marks look like injuries. The agency advises explaining where they came from to health care providers, so that they are not mistaken for signs of physical abuse. If you have children in school, elderly relatives in care, or an appointment in an emergency department, that sentence is the one to remember.
What the trial literature contains
In February 2012 a group at Beijing University of Chinese Medicine published what remains the fullest audit of cupping’s trial record, in the open-access journal PLoS One. Huijuan Cao and colleagues searched six databases, four of them Chinese, screened 1,294 citations, and assembled 135 randomized controlled trials published from 1992 through 2010. More than half appeared in the last three years of that window.
Three numbers from that paper describe the state of the evidence better than any adjective.
Of the 135 trials, 132 were published in Chinese, including three unpublished dissertations. Three were published in English.
Of the 135 trials, 133 reported that cupping, alone or combined with something else, was significantly effective for the condition studied.
Of the 135 trials, 4 reported any blinding method at all. That is 2.96 percent. Of those four, only two said who had been blinded, and in both cases it was the people assessing outcomes rather than the patients.
This is a property of a literature, not a verdict on a practice, and not a comment about a country. Chinese research output on traditional therapies is enormous, which is why a review of cupping finds mostly Chinese trials. A field in which 133 of 135 studies come out positive is a field where something other than the treatment is driving the result, and the usual suspects are design and reporting.
Cao’s quality table fills in the rest. Adequate randomization was described in 25 trials, 18.52 percent. Allocation concealment, which stops the person enrolling patients from knowing which group comes next, was adequate in 3. Three trials calculated a sample size in advance. Five reported how many participants dropped out, and none analysed their data on an intention-to-treat basis. Not one of the 135 trials was rated low risk of bias, and 84.44 percent were rated high risk.
Then there is the measuring instrument. An efficacy standard was reported in 126 trials, but 110 of them graded results on a composite four-step scale: cured, markedly effective, effective, ineffective. The reviewers noted that those categories are not internationally recognized and their exact meaning is open to interpretation. A trial that reports a cure rate without defining cure has reported a number, not a result.
Cao’s team did pool results where trials were similar enough. Wet cupping beat drug treatment on the number of patients cured of shingles, with a risk ratio of 2.07 and a confidence interval of 1.77 to 2.43 across five trials. For acne, wet cupping against medications produced a risk ratio of 2.14, with an interval of 1.42 to 3.22. Those figures are real, and they are not the same kind of object as a comparable number from a registered double-blind trial. The six conditions studied most often were shingles, Bell’s palsy, cough and breathlessness, acne, lumbar disc herniation and cervical spondylosis. The sports recovery and general muscle soreness uses that fill Western marketing barely appear.
A funnel plot of 39 trials showed potential asymmetry, which is the standard signature of publication bias, and the authors added that because they had excluded unpublished studies their review itself was likely affected by it.
Thirteen years and seventy-two more trials
The obvious hope would be that things improved. In April 2025 the same research center at Beijing University of Chinese Medicine published an update on cupping for pain, searching seven databases from January 2014 to January 2023 and including 72 trials with 5,720 participants between them.
Their finding on quality was a single sentence: “All included trials were assessed as having high risk of bias.”
Most reported benefit, with average risk ratios above 1.15 on cure rate and average pain reductions on a visual scale ranging from 0.16 to 7.0 centimeters. That range is worth staring at, because on a ten-centimeter scale 0.16 is nothing and 7.0 is implausible, and both came out of the same body of work. The authors acknowledged that their conclusions rest on low quality evidence.
The 2025 review draws no comparison with the 2012 audit, so the comparison here is this article’s own. Thirteen years and 72 more trials separate two reviews from the same Beijing centre, and neither found a single trial at low risk of bias. That the harsher of the two verdicts comes from a Chinese university of traditional medicine, with every professional incentive to find otherwise, is the most useful fact about this evidence base.
Newer condition-specific reviews follow the same shape. A 2026 meta-analysis of cupping for chronic neck pain pooled nine randomized trials with 401 participants and found a large pain effect, Cohen’s d of 1.17, with a confidence interval running from 0.26 to 2.09. An interval that wide is not a stable estimate. The functional disability result was tighter, 0.77 with an interval of 0.63 to 0.91, and the authors themselves noted methodological limitations in several of the included trials. A 2023 review of cupping for obesity, pooling 21 randomized trials, reported improvements in body weight, body mass index and waist and hip circumference, and no significant change in the waist-to-hip ratio or in body fat percentage, both of which were rated very low certainty. Soft measures moved and hard measures did not.
An overview of five earlier systematic reviews, published in 2011 by Korean researchers writing with Edzard Ernst, reached the summary that still fits: relatively clear evidence emerged for one indication only, that cupping may reduce pain, and even there doubts remain.
The blinding problem here is physical
Cupping leaves circular bruises that last for days. A participant can see them. So can anyone assessing the outcome. That makes a credible dummy treatment genuinely hard to build, rather than merely inconvenient. Either the sham leaves no marks, in which case anyone who looks at their own back knows what they got, or it leaves marks, in which case it is applying real suction and is not inert.
One attempt exists. In 2010, Korean researchers described a sham cupping device made by drilling a small hole in the cup so that negative pressure could not be maintained after suction. They tested it on 34 healthy volunteers, 17 receiving real cupping and 17 the sham, told them they would get one or the other, and asked them to guess. The two groups reported similar sensations, with a tendency for real cupping to feel stronger, 48.9 against 33.3 on a 100 millimeter scale.
The authors called their own study a pilot and wrote that more rigorous research was warranted. It has not appeared. The study used healthy volunteers rather than patients, and it measured sensation rather than marks. A device producing little or no negative pressure produces little or no bruise, so the central problem is exactly the one the paper did not address.
Cao’s team, writing from inside the field, reached the pragmatic conclusion that blinding patients may be out of reach and that researchers should at minimum blind the people assessing outcomes. That recommendation sits in the same paper as the count behind it: of those 135 trials, 4 mentioned blinding at all. The deeper argument about what a sham control can and cannot establish for a hands-on therapy is set out in this library’s article on acupuncture.
Nobody in the united states licenses cupping
There is no cupping license in any American state, no national cupping certification board comparable to the certifying bodies that exist for other hands-on practices, and no cupping entry among the programmatic accrediting agencies recognized by the US Department of Education. That register covers programs in osteopathic medicine, naturopathy, chiropractic, acupuncture and massage therapy. Cupping appears nowhere on it.
In practice, cupping is performed under somebody else’s license: by licensed acupuncturists, massage therapists, physical therapists, chiropractors and physicians, each governed by their own board’s scope of practice. Which means the scope question has to be asked of that board, in that state, about that profession.
The question matters most for wet cupping, because wet cupping deliberately breaks the skin to withdraw blood. That is a procedure most state licensing frameworks were not written with in mind. A consumer cannot assume that a practitioner’s license authorizes it, and a practitioner who offers it should be able to say which provision of their scope of practice covers it.
Cupping is also not exclusively Chinese, and coverage that treats it as a Chinese practice is simply wrong on the history. A 2017 narrative review by researchers at Saudi Arabia’s National Center of Complementary and Alternative Medicine, which screened 625 articles and included 83, records documented use in early Egyptian and Chinese medical practice, with contributions from many civilisations, and notes that the practice fell out of favor in the seventeenth and mid-eighteenth centuries before recovering. The same review found inconsistent historical data about origins, definitions, instruments and procedures. In the Islamic tradition the practice is hijama and carries prophetic-medicine endorsement, which is a fact about the tradition rather than a claim about effect.
In the traditional Chinese framing, cups are placed on acupuncture points along meridian lines, and the proposed mechanism is that suction produces local blood pooling that in turn produces a therapeutic effect. Cao’s team, reporting that framework, also wrote plainly that the mechanism of cupping therapy is not clear.
Five questions before a session
These are the questions that follow from the evidence above rather than from general advice about picking a practitioner.
- Dry or wet? Ask it first and ask it plainly, because the answer changes everything else on this list.
- If wet, what license are you working under, and what in its scope of practice covers breaking skin to withdraw blood? A clear answer is reassuring. A vague one is informative.
- What is single-use here and what gets reused, and how is reused equipment sterilized between clients? Bloodborne transmission through shared unsterilized equipment is the specific risk the federal agency names.
- How many sessions are you proposing, over what period? Repeated blood loss is the documented route to anemia, and the documented cases involved repetition.
- What are you treating, and what happens if it does not improve? A practitioner who has an answer to the second half of that question is working inside their limits.
What to carry away
Dry or wet is the dividing line for evidence and for risk, and no other distinction in this subject matters as much. If it is dry cupping, the risk picture is mostly marks, occasional burns from flame techniques and skin irritation, and the realistic expectation is a short-term change in how an area feels, drawn from a literature in which not one trial across two major reviews covering 207 studies was rated low risk of bias. If it is wet cupping, you are consenting to a procedure that removes blood, and the strongest single piece of evidence about it in a mainstream medical journal is a list of causes of anemia.
Tell the clinicians who look after you that the marks are cupping marks. And treat whatever you were hoping cupping would fix as something that still needs a diagnosis. Unexplained fatigue, breathlessness on mild exertion, unusual pallor, blood in the stool, a new lump, unintended weight loss, or pain that is worse week on week all belong with a doctor. Anemia in particular needs its cause found rather than assumed, because the thing that turned up most often in that Korean study was not cupping.
Sources
- PLoS One. “Cao H, Li X, Liu J. "An updated review of the efficacy of cupping therapy." PLoS One, 2012 (135 RCTs; 132 in Chinese; 133 reporting benefit; blinding reported in 4). PubMed. DOI: 10.1371/journal.pone.0031793.” pubmed.ncbi.nlm.nih.gov
- Journal of Traditional Chinese Medicine. “Wang Y, Dong S, Li B, Han M, Cao H. "Update evidence of effectiveness on pain relieving of cupping therapy: a systematic review and meta-analysis of randomized controlled trials." J Tradit Chin Med, 2025 (72 trials, 5,720 participants). PubMed. DOI: 10.19852/j.cnki.jtcm.2025.02.002.” pubmed.ncbi.nlm.nih.gov
- Internal Medicine (Tokyo). “Yun GW, Yang YJ, Song IC, et al. "A prospective evaluation of adult men with iron-deficiency anemia in Korea." Intern Med, 2011 (206 men; bloodletting cupping the probable cause in 11, or 5.3 percent). PubMed. DOI: 10.2169/internalmedicine.50.5289.” pubmed.ncbi.nlm.nih.gov
- Evidence-Based Complementary and Alternative Medicine. “Xu S, Wang L, Cooper E, Zhang M, Manheimer E, Berman B, Shen X, Lao L. "Adverse events of acupuncture: a systematic review of case reports." Evid Based Complement Alternat Med, 2013. PubMed. DOI: 10.1155/2013/581203.” pubmed.ncbi.nlm.nih.gov
- Acupuncture in Medicine. “Lee MS, Kim JI, Kong JC, Lee DH, Shin BC. "Developing and validating a sham cupping device." Acupunct Med, 2010 (34 healthy volunteers; authors describe it as a pilot). PubMed. DOI: 10.1136/aim.2010.002329.” pubmed.ncbi.nlm.nih.gov
- Journal of Acupuncture and Meridian Studies. “Lee MS, Kim JI, Ernst E. "Is cupping an effective treatment? An overview of systematic reviews." J Acupunct Meridian Stud, 2011. PubMed. DOI: 10.1016/S2005-2901(11)60001-0.” pubmed.ncbi.nlm.nih.gov
- Journal of Back and Musculoskeletal Rehabilitation. “Torad AA, Taha MM, Elbosaty RF, et al. "Effect of cupping therapy on chronic neck pain: a systematic review and meta-analysis of randomized controlled trials." J Back Musculoskelet Rehabil, 2026 (9 RCTs, 401 participants). PubMed. DOI: 10.1177/10538127261446411.” pubmed.ncbi.nlm.nih.gov
- Medicine (Baltimore). “Kang D, Shin WC, Kim T, et al. "Systematic review and meta-analysis of the anti-obesity effect of cupping therapy." Medicine, 2023 (21 RCTs). PubMed. DOI: 10.1097/MD.0000000000034039.” pubmed.ncbi.nlm.nih.gov
- Journal of Integrative Medicine. “Qureshi NA, Ali GI, Abushanab TS, et al. "History of cupping (Hijama): a narrative review of literature." J Integr Med, 2017 (authors at Saudi Arabia's National Center of Complementary and Alternative Medicine). PubMed. DOI: 10.1016/S2095-4964(17)60339-X.” pubmed.ncbi.nlm.nih.gov
- National Center for Complementary and Integrative Health, National Institutes of Health. “Cupping (content date: November 2018).” nccih.nih.gov
- US Department of Education. “Programmatic Accrediting Agencies.” ed.gov