Injury epidemiology in MMA, read properly

1. The number everyone quotes, and what is inside it

Start with the meta-analysis, because almost every popular claim about MMA injury risk traces back to it.

Lystad, Gregory and Wilson published a systematic review and meta-analysis in Orthopaedic Journal of Sports Medicine in 2014. They pooled six studies of injury in professional, sanctioned mixed martial arts competition and produced a random-effects summary estimate of the injury incidence rate per athlete-exposure. Their own sentence: "The IIRAE summary estimate was found to be 228.7 (95% CI, 110.4-473.5)."

One athlete-exposure is one fighter in one bout. So the point estimate says that across these six datasets, roughly 229 injuries were recorded for every 1,000 fighter-appearances — a shade under one recorded injury for every four fighters who walked out.

Now the qualifications, and they are larger than the number.

The heterogeneity statistic is I² = 97.1%. That is an estimate of how much of the variation between the six studies is real difference rather than sampling noise, and 97.1% is about as high as that statistic goes. The six studies are not six samples of one quantity. They are six different quantities.

The confidence interval runs from 110.4 to 473.5. The upper bound is more than four times the lower. Any sentence that prints 228.7 without that interval has converted a wide, unstable pooled estimate into a fact, and the review itself would not support it.

And then the mechanism. The review states that "in all but 1 study what constituted a reportable injury was at the discretion of the attending ringside physician." Not a written case definition. Not a threshold agreed in advance. A working doctor at a cageside desk, deciding in real time and under the pressures of that desk, what was worth writing on a form. Six commissions, six eras, six professional cultures about what counts.

The same review is explicit elsewhere that "the operational injury definition in all MMA studies to date is unclear."

That is the state of the primary evidence. It is not nothing — it is a real, consistently gathered administrative record of a real sport — but it is a record of what ringside physicians wrote down, and the rate it produces is as much a property of the recording as of the fighting.

2. The commission numbers, one at a time

Underneath the pooled estimate are individual studies, and they are more useful than the pool because each one carries a single, identifiable recording system.

Ngai, Levy and Hsu reviewed 635 professional MMA matches sanctioned by the Nevada Athletic Commission between March 2002 and September 2007 — 1,270 fight exposures. Injuries were those documented in commission medical records. The rate was 23.6 injuries per 100 fight participations, competition only. Three hundred of the 1,270 athletes were recorded as injured.

Bledsoe and colleagues had looked at an overlapping Nevada window earlier: 171 professional matches, 220 fighters, September 2001 to December 2004, again on commission medical and outcome records. That produced 28.6 injuries per 100 fight participations, or 12.5 per 100 competitor rounds, and 69 of the 171 matches — 40.3% — ended with at least one fighter recorded injured.

Those two studies share a commission, a definition type and a decade, and they agree reasonably well. That is the point. Fix the recording system and the numbers stabilise.

Widen the frame and they do not. A 2025 systematic review of injuries after adoption of the Unified Rules assembled 43 reports and found competition injury rates ranging from 23.6 to 54.5 per 100 athlete-exposures. The review states plainly that the reports "varied widely in design, injury definitions, and data collection methods." That range is not a distribution of one measured quantity across contexts. It is a set of answers to different questions laid side by side.

One more, because it is the figure most often misused. A 2018 systematic review reported a weighted average of 246.4 injuries per 1,000 athletic encounters for 2,407 males, and 101.9 per 1,000 for the single study of 108 females. Competition, mixed professional and amateur, ringside records. The female figure rests on one study of 108 women. It is the only thing there is, and it is not a stable estimate of anything.

Within that same review, one study separated by professional status: 135.5 per 1,000 athlete-exposures for professionals against 71.0 for amateurs, competition, ringside-record definition. Two later reviews report the same direction qualitatively. One study is not a finding about amateur MMA, and the field has not followed it up.

3. The experiment that measured the definition

Everything above is subject to one objection: nobody knows how much the ringside record misses. In 2026 that stopped being a rhetorical question.

Doherty, Barley and Fortington ran two linked studies on Western Australian combat sports. The first surveyed 175 Australian MMA and Muay Thai competitors across 26 events, collecting athlete self-reports seven days after competition using the OSTRC-H2 instrument and the IOC health-problem definition — that is, anything that reduced the athlete's health, irrespective of whether medical attention was sought or training was lost. Eighty-one competitors reported 411 health problems, 315 of them injuries and 96 illnesses. Thirty-nine per cent of respondents reported at least one. Expressed against competition exposure time, that came to 20.1 health problems per 100 minutes of competition exposure for MMA (95% CI 16.5–24.4), against 25 per 100 minutes for Muay Thai. Median days impacted for MMA: 20.

The second study did the thing nobody had done. Across 29 Western Australian events in 2022 and 2023, with 250 questionnaire responses, it compared the ringside physician's record against the athlete's own report of the same competition.

Physicians recorded an injury for 25 of the 250 respondents — 10%. Athletes reported 321 health problems among 97 of them — 39%. Agreement between the two methods was 15%. Of the 23 time-loss injury-related health problems the athletes reported, the physicians had recorded 5; they missed 78%. The authors conclude that "neither ringside medical reports nor athlete self-reports alone fully capture competition injuries, with only 15 % agreement between methods," and estimate that pairing them could capture up to 4.6 times more time-loss injuries.

Hold that against the meta-analysis. A pooled estimate built entirely out of ringside records, with a fourfold confidence interval, sits on top of a data source that in a direct comparison failed to record more than three quarters of the time-loss problems the athletes themselves reported a week later.

This is a 29-event study in one Australian state, and it should not be extrapolated into a correction factor for Nevada in 2004. But it establishes the direction and rough size of the gap, and it is the strongest empirical demonstration available that the definition and the data source, not the sport, drive the rate. The same logic applies to a fighter's own history, which is the argument made at length in what an injury log is actually for.

4. The standard that already exists

There is a published methodology consensus, and MMA has largely not used it.

The IOC consensus statement on recording and reporting epidemiological data on injury and illness in sport, published in 2020 in British Journal of Sports Medicine and co-published in Orthopaedic Journal of Sports Medicine, sets out the definitions the field is supposed to share. It remains the current version; a search on 2026-09-22 found no successor, and 2025 and 2026 papers still cite it as current, though no IOC page was fetched confirming it has not been superseded.

The statement defines a health problem as "any condition that reduces an athlete's normal state of full health, irrespective of its consequences on the athlete's sports participation or performance or whether the athlete sought medical attention." That is the widest of the categories, and it is deliberately wider than anything a ringside form captures.

It defines an injury as "tissue damage or other derangement of normal physical function due to participation in sports, resulting from rapid or repetitive transfer of kinetic energy," and an illness as "a complaint or disorder experienced by an athlete, not related to the injury," including physical, mental and social well-being problems.

Inside the health-problem category it nests two narrower ones. A medical-attention health problem is one "that results in an athlete receiving medical attention." A time-loss health problem is one "that results in a player being unable to complete the current or future training session or competition."

Three nested definitions, in a fixed relationship: every time-loss problem is a health problem, but not every health problem produces time loss, and the ratio between them is not a constant. That is why the Australian work found 39% of athletes reporting something and 10% appearing in a physician's record. Both counts were correct. They were counting different objects.

The statement is also explicit about the denominator. It says that "incidence-based measures that provide a standard time window for the population at risk (injuries per hour) are preferable ... because time-based measures better facilitate comparison across sports." Per-hour, not per-exposure. Almost the entire MMA competition literature is per-athlete-exposure, where one exposure is a bout of at most about fifteen to twenty-five minutes of scheduled time and frequently far less. That is exactly the measure the consensus identifies as harder to compare across sports, because the time at risk varies from one exposure to the next.

A 2025 review's own summary of where MMA stands is this: "Analysis of current literature emphasizes a lack of standardized definitions, data on training injuries, and female injuries."

5. What gets injured in a fight

With the caveats established, the competition body-region data is consistent enough to be worth stating, because it is the part of the picture the different recording systems mostly agree on.

Across the six pooled studies, the head accounted for 66.8% to 78.0% of recorded competition injuries. The wrist and hand accounted for 6.0% to 12.0%, the upper limb overall 11.8% to 22.8%, and the lower limb 4.0% to 29.9%. The 2025 post-Unified-Rules review reports head and neck at 29.5% to 75.9% across its more heterogeneous set of reports.

By type, the pooled ranges were: laceration and abrasion 36.7% to 59.4%, fracture 7.4% to 43.3%, concussion or nerve injury 3.8% to 20.4%, and contusion 1.8% to 3.3%. The 2025 review reports soft tissue injury — lacerations, abrasions and contusions together — at 20.7% to 56.9%.

That contusion figure is the interesting one. The review flags 1.8–3.3% as implausibly low, and reads it as direct evidence of under-recording at ringside. Fighters finish bouts covered in bruises. A recording system in which bruising is under 4% of injuries is a system that is not recording bruises, which tells you what else it might not be recording.

The upper limb picture is better resolved in a single-commission study. Across 408 UFC fights sanctioned in Nevada between 2016 and 2019, there were 81 upper limb injuries — a rate of 9.9 per 100 athlete-exposures, with the hand alone at 6.61 per 100 (p<0.001 against other upper-limb sites). The commonest recorded mechanism was striking the opponent. Decision finishes carried the highest upper-limb rate at 12.6 per 100 athlete-exposures, though the difference across finish types did not reach significance (p=0.115). The highest divisional rates were women's strawweight at 20.5, women's flyweight at 19.2 and men's flyweight at 13.8 per 100 athlete-exposures. The female rate was numerically higher than the male but not significantly so (p=0.454).

The hand result is worth sitting with, because it is one of the few places where a competition rate has an obvious training analogue — the same structures take the same loading every time a fighter hits pads or a bag, which is the subject of hand and wrist injuries in striking.

6. Getting stopped, and losing

Two variables predict competition injury more strongly than anything else measured, and neither is a characteristic of the fighter.

Pooled across the professional competition studies, KO/TKO bouts produced 2.37 times the injuries of submission bouts (95% CI 2.01–2.82), and losing fighters sustained 2.99 times the injuries of winners (95% CI 2.26–3.95). Both intervals are tight enough to take seriously, which distinguishes them from the pooled incidence estimate.

Stoppage-type rates compiled in the 2018 review run: KO/TKO 173.9 per 1,000 athlete-exposures in males and 175.9 in females; submission 228.6 per 1,000; referee's decision 98.2 per 1,000. Those four numbers come from different contributing studies and their comparability with each other was not established, so read them as four separate observations rather than a ranking.

The losing effect reappears in every dataset that looks for it. In 2,208 professional MMA fights in Nevada between 2001 and 2020, 62.8% of fighters who sustained an eye injury lost the bout. In a cohort of 1,462 UFC fighters in Nevada between 2010 and 2020, losing was the strongest predictor of facial injury (p<0.001), alongside male sex (p=0.026), heavier weight (p=0.028), more rounds (p=0.019) and a non-submission finish (p=0.017).

None of this is a finding about fragility. It is largely a finding about how bouts end: the fighter who is being stopped by strikes is by definition the one absorbing them, and a record generated at the moment of stoppage will record that.

7. The head and the face, in detail

The head is where the competition literature is densest, and it is worth separating what each study actually measured.

In 408 UFC fights sanctioned by the Nevada State Athletic Commission between 2016 and 2019 — 816 athlete-exposures — there were 288 head injuries. The head injury rate was 35 per 100 athlete-exposures, and traumatic brain injury specifically 16 per 100, significantly more than fractures (p=0.003). Males 37, females 23 per 100 athlete-exposures. KO/TKO was the highest-rate finish (p<0.001), and rates rose with weight division. The injury definition here was the ringside physician report supplemented by play-by-play video review — a broader ascertainment than record review alone, which is part of why the rate is higher.

Facial injury in the 1,462-fighter Nevada UFC cohort, 91.0% male, ran at 15.8% of fighters — lacerations 12.0%, fractures 3.6% — on NSAC medical injury reports.

Maxillofacial trauma has been catalogued separately: 319 injuries in 259 UFC fighters between 2015 and 2019, 291 (91.2%) in men and 28 (8.8%) in women, from ringside physician reports gathered across 16 US states and 24 countries. The longest mean medical suspensions in men followed light heavyweight (110.68 days) and welterweight (108.25 days) bouts; the longest in women followed strawweight bouts (88.92 days). A medical suspension is an administrative decision by a commission, not a clinical recovery measure and not a statement about healing time, and it should not be read as one.

The eye data is the most complete of all, for the plain reason that eye injuries are visible and consequential. Across 2,208 professional MMA fights in Nevada from 2001 to 2020 there were 369 eye injuries in 363 fighters, with yearly rates between 2.56 and 12.22 eye injuries per 100 fighters. Forty-three fighters received no-contact requirements averaging 8.9 weeks, with a range of 1 to 24 weeks.

Note what 369 injuries in 363 fighters does to the arithmetic: injuries and injured people are close but not equal, and in other datasets they are much further apart — 96 injuries in 78 fighters in the early Nevada cohort. This is why a per-exposure injury rate cannot be read as a percentage of fighters. It is a count of events over a count of appearances, and the two can exceed each other in either direction.

8. Three concussion numbers that are not the same number

If one section of this article is worth reading twice, it is this one, because these three figures circulate interchangeably and describe entirely different things.

A commission-record incidence. The Nevada cohort of 635 professional matches reported severe concussion at 15.4 per 1,000 athlete-exposures, about 3% of all matches, with "severe concussion" as recorded in commission records. Note the unit switch — that same paper's overall injury rate is expressed per 100 fight participations. The 2025 post-Unified-Rules review reports concussion rates of 14.7 to 16.1 per 100 athlete-exposures across reports with non-standard definitions, which is a different unit again and sits on a heterogeneous definition base.

A stoppage rate from video. Across 143 Australian MMA events between 2020 and 2023, KO/TKO secondary to head strikes ran at 16.6 per 100 athlete-exposures in amateur men and 18.7 in professional men; 12.6 in amateur women and 7.4 in professional women. Thirty-four per cent of male bouts ended by head-strike KO/TKO against 23% of female bouts (p=0.01). These figures were derived by reviewing fight outcomes on video. A stoppage from head strikes is a stoppage. It is not a diagnosed concussion, and treating it as one inflates a diagnosis rate out of an officiating decision.

A pooled prevalence. A 2026 meta-analysis reports pooled concussion prevalence of 20.68% (95% CI 17.66–23.69). Prevalence is a proportion of athletes reporting the condition over a career or a period, not an incidence rate, and it cannot be compared with any per-exposure figure on this page. Its age subgroups — 20.37% (95% CI 14.82–25.91) for athletes aged 31–40 against 19.50% (95% CI 3.72–35.29) for 20–30 — have intervals that overlap almost completely, and the younger group's interval spans nearly the whole plausible range.

Three legitimate numbers. Three incompatible denominators. Anyone arranging them into a trend is producing an artefact of arithmetic.

What happens after a fighter is stopped is governed by clinical assessment and by commission suspension rules, not by any of these rates; the jurisdictional side of that is covered in returning to training after a knockout.

9. Training: the missing nine-tenths

Every figure so far describes competition. Competition is a tiny fraction of the exposure.

A fighter who takes three bouts in a year accumulates perhaps forty-five minutes of scheduled competition time against something on the order of five hundred training hours. The competition literature is the part that has been measured, not the part where the exposure is.

The 2018 systematic review put it without hedging: "There are no studies of training injuries of professionals or injuries of amateurs or long-term follow-up of musculoskeletal injuries or neurological damage." The 2025 review repeated that standardised training-injury data is still lacking. Nothing located on 2026-09-22 changes that. No prospective training-injury surveillance of professional MMA fighters exists.

The nearest thing is a 2026 Austrian study, and its design must travel with its number. It is a retrospective cross-sectional survey of 112 MMA athletes — 104 men and 8 women, 59% competitive — recalling 24 months of injuries across 93,857 total exposure hours, with data collected between April and December 2023. The injury definition is any physical complaint irrespective of medical attention or time loss, which is the broadest IOC category. The exposure is predominantly training.

It reports 1.4 injuries per 1,000 hours overall, 1.03 per 1,000 hours for competitive athletes and 1.56 per 1,000 hours for non-competitive athletes. The non-competitive group sustained significantly more total injuries (p=0.003), while the odds of a severe injury were lower in that group (OR 0.55, p=0.042).

It also gives the body-region distribution, and this is where the training picture separates from the competition picture completely. Of 307 affected sites: head and neck 19.8%, ankle and foot 15.3%, hand and fingers 14.0%, knee and calf 12.3%, shoulder girdle and upper arm 12.3%. By type: joint sprains 21.2%, ligament strains 17.3%, muscle strains 15.6%, fractures 10.1%, traumatic brain injury 5.2%.

Set that beside 66.8–78.0% head in competition. "The head is the most injured body part in MMA" is true of sanctioned competition and materially false of training exposure. Those two distributions describe different activities and must never be merged into one claim about the sport.

Two structural caveats on the training number. The competitive and non-competitive athletes are pooled into one exposure pool, so the 1.4 figure is not a fighter's rate. And 8 women out of 112 is not a basis for any statement about women's training injuries.

10. Women, amateurs, adolescents

Population limits in this literature are severe, and they are stated here rather than scaled.

Women. The 2018 review's sex split was 108 women against 5,374 men. The training survey had 8 women out of 112. Maxillofacial injuries: 8.8% female. The UFC facial-injury cohort: 9.0% female. Where sex-stratified rates exist they come from single commissions or single countries: head injury 23 versus 37 per 100 athlete-exposures (women versus men) in the Nevada UFC cohort; head-strike KO/TKO 12.6 in amateur women and 7.4 in professional women against 16.6 and 18.7 in men in the Australian video review; upper-limb rate numerically higher in women but not significant. None of that is a general statement about injury risk in women's MMA, and sex differences in bone, energy availability and concussion are pathway differences rather than multipliers — they do not become a coefficient to apply to a male rate.

Amateurs. One study within one review separates amateurs, at 71.0 per 1,000 athlete-exposures. The Australian video work provides amateur stoppage rates. No amateur-specific body-region or severity distribution was located. For a sport whose participation is overwhelmingly amateur, this is a near-total absence.

Adolescents. Sanctioned adolescent MMA has no injury surveillance at all. What exists is emergency-department data covering all combat sports and martial arts. A 2026 study found that adolescents aged 10 to 19 accounted for over 60% of an estimated 186,471 craniofacial injuries between 2014 and 2023, 87% male, with lacerations 38.5%, concussion or internal head injury 24.1% and fractures 10.7%, and hospital admission around 2–3%. That is a description of who arrives at emergency departments. It has no participation denominator, so it cannot produce a rate and it cannot be read as adolescent MMA competition risk.

Emergency-department data has the same limit wherever it appears. A NEISS-based study of 39,181 estimated ED-presenting injuries across BJJ, judo and MMA between 2008 and 2015 found that the majority of MMA presentations arose in competition while most BJJ and judo presentations arose in non-competitive grappling, with abrasions and contusions the commonest MMA diagnosis and the head the most injured region. ED presentation is a fourth definition, narrower than all three IOC categories: it counts injuries severe enough, or frightening enough, to send someone to a hospital. A broader martial-arts NEISS study covering 2015 to 2024 estimated 277,753 injuries from 7,613 cases, with women showing a higher proportion of youth injuries (46.1% versus 33.1%) and of foot, hand and leg injuries and strains and sprains, and male incidence rising after 2020 particularly for grappling and MMA. Those are shares of presentations, not rates.

11. Comparisons that hold, and comparisons that do not

Cross-sport comparison is where this subject goes wrong most often, so the rule used here is simple: a comparison is admissible only when both sports were counted by the same recording system.

Two are.

Kickboxing, through the same commission. A 15-year retrospective cohort of Nevada Athletic Commission records covering 481 kickboxing fighters across 57 events, 976 athlete-exposures and 9,562 exposure minutes reports 390.1 injuries per 1,000 athlete-exposures (95% CI 351.9–431.4) and 39.7 per 1,000 minutes of exposure (95% CI 35.8–43.9). Head 57.8%, lower extremity 26.1%; laceration 70.6%, fracture 20.6%. Professionals 2.51 times amateurs (95% CI 1.39–4.55), losers 3.48 times winners (95% CI 2.73–4.44). Because this is the same commission-record definition as the Nevada MMA studies, the comparison is admissible — and the direction of the professional/amateur and loser/winner effects is the same in both sports.

Boxing, on eye injury, twice. Texas Department of Licensing and Regulation post-match physical reports covering 3,070 participants between January 2019 and January 2022 give ocular injury incidence of 12.2 per 100 matches in MMA against 9.7 per 100 matches in boxing, with ocular injury associated with losing in boxing (p=0.011) but not in MMA (p=0.232). Nevada Athletic Commission records from 2000 to 2020 — 1,442 MMA injuries from 2,704 contests and 1,539 boxing injuries from 4,313 contests — give eye trauma as 25.59% of MMA injuries against 47.63% of boxing injuries, orbital fractures as 17.62% of MMA eye injuries against 3.14% in boxing, and odds of requiring physician evaluation after an eye injury 1.823 times higher in MMA (95% CI 1.408–2.359). Same commission, same recording system, both sports.

Those two results point in different directions depending on what is being measured — boxing has the larger share of injuries that are eye injuries, MMA the higher ocular injury rate per match and the greater proportion of orbital fractures — which is precisely why "MMA is safer than boxing" is not a claim this evidence supports. That phrase traces to a hypothesis in a 2006 paper covering 171 bouts, suggesting that lower knockout rates in MMA than boxing might help prevent brain injury. It was a hypothesis, not a comparative safety finding, and it has not become one.

What cannot be compared: MMA against American football, rugby or soccer. Those literatures are overwhelmingly time-loss-based and expressed per 1,000 player-hours. The MMA competition literature is ringside-record-based and expressed per 100 or per 1,000 athlete-exposures, where one exposure is a bout of at most about fifteen to twenty-five minutes. There is no conversion between those denominators, and no cross-sport ranking is printed here.

Nor is "MMA has the highest injury rate of any sport" supportable. The strongest sourced version of that claim, in a 2026 meta-analysis, is that the MMA injury rate "seems to be higher than most, if not all, other popular and common combat sports" — hedged, and restricted to combat sports, from a review that also reports injury prevalence across its included studies "ranging from 0.2 to 92.7". The same problem of incompatible denominators runs through grappling too, which is why the numbers in injury rates in BJJ and grappling cannot be laid against the ones on this page.

12. Severity is the part nobody measured

The 2014 review notes that none of its included studies reported injury severity. Twelve years later that is still close to true.

What the commission literature has instead is medical suspension days — the maxillofacial study's 110.68 days for light heavyweight men, 88.92 for strawweight women. A medical suspension is a commission's administrative decision about eligibility to be licensed for a contest. It reflects a regulatory posture, a jurisdiction's schedule and a physician's caution. It is not a clinical recovery measure, and two fighters with identical injuries in different states can receive different suspensions.

The only athlete-reported recovery duration located in this literature is the Australian self-report study's median of 20 days impacted for MMA competitors. One figure, from 26 events in one state.

The IOC consensus asks for severity to be recorded as days unavailable, from onset to full availability. Almost nothing in MMA does this. It is a straightforward thing for an individual athlete to record and an expensive thing for a research programme to collect at scale, which is roughly why the situation persists.

13. A worked scenario, and what it is not

This is arithmetic on published group rates. It is not a prediction about any individual, and Fighter Cut does not coach a roster — the athlete below is invented to make the arithmetic visible.

Take an amateur who takes three sanctioned bouts in a year. Three athlete-exposures.

At the commission-record definition — the narrow one, what the ringside doctor wrote down. Using the amateur figure of 71.0 injuries per 1,000 athlete-exposures from the single study that reports it: 3 × 0.071 = 0.21 recorded injuries expected across the year. Borrow the professional Nevada figure of 23.6 per 100 fight participations instead: 3 × 0.236 = 0.71. Same athlete, same year, same three fights. The expected count more than triples depending only on which cohort and record system the rate was taken from.

At the any-complaint definition — anything that hurt afterwards. The Australian self-report study found 39% of competitors reporting at least one health problem seven days after a bout, median 20 days impacted for MMA. Over three bouts, and treating the bouts as statistically independent purely for illustration — an assumption that is almost certainly wrong, since prior injury is itself a listed risk factor — the chance of no reported health problem all year is 0.61³ ≈ 0.23. That leaves roughly three chances in four of at least one post-fight health problem in a three-fight year, against an expectation of well under one recorded injury.

The gap between 0.21 and about 0.77 is not a disagreement about how dangerous MMA is. It is two different questions, answered correctly.

Training is the rest of the exposure. Ten training hours a week for fifty weeks is 500 hours. At the Austrian survey's 1.4 injuries per 1,000 hours, any-complaint definition: 500 × 0.0014 = 0.7 injuries expected from training in the year — comparable in raw count to the competition figure, from a completely different body-region profile: ankle and foot, hand and fingers, knee, shoulder rather than head and face. Printed on the same line, as it must be: that rate comes from a retrospective self-report survey of 112 athletes including 8 women in one country, and no study of professional training injuries exists.

None of these four numbers is a personal forecast. They are group rates multiplied by an exposure count, and an individual's outcome is not drawn from them. The reason to run the arithmetic at all is to see how far apart the answers sit when only the definition changes. Camp-level questions about how that exposure is distributed — how much of it is hard sparring, and when — are taken up in how much hard sparring a camp can carry.

14. What to do with all of this

The practical conclusion is narrow, and it is about reading rather than training.

When a number about MMA injury risk appears — in a broadcast, a gym conversation, a news story — four questions settle whether it means anything. What counted as an injury? What was the denominator? Which cohort, in which jurisdiction, in which years? Competition or training? A figure that cannot answer all four is not comparable to any other figure, including the one it is being compared with.

Most of the numbers that circulate fail on the first question and the fourth. "MMA injury rate is 23%" fails on both: it is a folk conversion of a per-exposure rate into a proportion of people, and it is a competition figure being used to describe a sport that is mostly training.

For an individual athlete, the honest version of this literature is that it describes competition days in a few commissions, mostly for professional men, mostly in North America and Australia, using a record that in direct comparison missed most of what athletes said had happened to them. It says something real about the shape of competition injury — head and face dominant, hands next, losing and getting stopped the strongest correlates. It says almost nothing about the training year, and nothing at all about adolescents in sanctioned MMA.

The one thing an athlete can do that the literature cannot is fix a definition to their own history and keep it consistent. That does not produce a rate, and it does not diagnose anything. It produces a record where there would otherwise be recall, which is the least stable definition of all. The general arguments for that sit in the fight camp pillar.

What we could not verify

No prospective training-injury surveillance of professional MMA fighters exists. The 2018 review said so and the 2025 review repeated it. The only study with real training-hour denominators is a retrospective cross-sectional survey of 112 Austrian athletes with 24-month recall, 8 of them women, pooling competitive and non-competitive athletes into one exposure pool. Every training figure on this page comes from that one study.

No MMA competition study applies the IOC health-problem framework prospectively across a full season. The two Australian studies are the methodological state of the art and cover 26 and 29 events in one state.

"Around 23% of MMA fighters are injured." Refused. It is a conversion of 228.7 injuries per 1,000 athlete-exposures into a proportion of people. The denominator is exposures and the numerator is injuries; fighters sustain more than one injury in a bout (96 injuries to 78 fighters in one Nevada cohort, 369 eye injuries to 363 fighters in another); and several studies report rates above 100 per 100 athlete-exposures, which no percentage can express. The related "your chance of injury per fight is 23.6%" is refused for the same reason — the coincidence between that rate and the separately reported 300-of-1,270 injured athletes in the same paper does not license the conversion anywhere else.

"MMA is safer than boxing." Refused. Traced to a hypothesis in a 2006 paper covering 171 bouts about knockout rates and brain injury. The actual head-to-head evidence, from two shared recording systems, is mixed: boxing has the higher share of injuries that are eye injuries, MMA the higher ocular injury rate per match and the higher proportion of orbital fractures. The specific comparisons are reported above; the blanket claim is not.

"85–90% of MMA fighters are injured during a training camp." Chased to origin on 2026-09-22 and not found. It is absent from the 2014 meta-analysis, the 2018 review, the 2025 review, the 2026 meta-analysis and the 2026 Austrian survey, and from every indexed study located. The nearest real figure is 127 injuries in 75 of 112 athletes over 24 months and 93,857 hours at the broadest possible definition — a two-year figure, not a per-camp one. There is no camp-level denominator in this literature at all.

Cross-sport comparisons with football, rugby or soccer. Refused. Those literatures use time-loss definitions and per-1,000-player-hour denominators; the MMA competition literature uses ringside records and per-exposure denominators. No conversion exists. Likewise "MMA has the highest injury rate of any sport" — the sourced version is hedged and restricted to combat sports.

Whether a post-2020 IOC consensus statement exists. A search on 2026-09-22 returned none, and 2025 and 2026 papers still cite the 2020 statement as current. No IOC page was fetched confirming it has not been superseded, so this is an absence of search evidence rather than a positive statement. The related OSIICS injury classification system has moved to Version 15 (2024), but that is a coding system, not a replacement consensus.

How much ringside recording misses, for injury types other than time-loss health problems. Quantified at 78% for time-loss problems in one Australian dataset, and inferred from implausibly low contusion counts in the pooled competition data. The multiplier for other injury types is unknown.

Long-term neurological outcome data for MMA cohorts is absent from this literature. The 2018 review names the absence explicitly. Nothing located establishes career or post-career outcomes in MMA cohorts.

Geography. The competition literature is dominated by the Nevada Athletic Commission, with Australian and Texan data added since 2021. Nothing was located from Europe, Asia or South America at commission-record quality.

Sources

Sourced to

1. The Epidemiology of Injuries in Mixed Martial Arts: A Systematic Review and Meta-analysis — Lystad RP, Gregory K, Wilson J, Orthopaedic Journal of Sports Medicine, 2014;2(1):2325967113518492. DOI 10.1177/2325967113518492, PMCID PMC4555522
2. Injury trends in sanctioned mixed martial arts competition: a 5-year review from 2002 to 2007 — Ngai KM, Levy F, Hsu EB, British Journal of Sports Medicine, 2008;42(8):686–689. DOI 10.1136/bjsm.2007.044891, PMID 18308883
3. Incidence of injury in professional mixed martial arts competitions — Bledsoe GH, Hsu EB, Grabowski JG, Brill JD, Li G, Journal of Sports Science and Medicine, 2006;5(CSSI):136–142. PMID 24357986, PMCID PMC3863915
4. Systematic review of injuries in mixed martial arts — Thomas RE, Thomas BC, The Physician and Sportsmedicine, 2018;46(2):155–167. DOI 10.1080/00913847.2018.1430451, PMID 29347856
5. Injuries in Mixed Martial Arts After Adoption of the Unified Rules of MMA: A Systematic Review — Zachovajevas V, Engebretsen L, Moatshe G, Zachovajevas P, Røise O, Orthopaedic Journal of Sports Medicine, 2025;13(7). DOI 10.1177/23259671251342578, PMID 40620723
6. A systematic evaluation of musculoskeletal injuries in mixed martial arts athletes: a meta-analysis — Huang Y, Si W, Xue Y, Yin W, Journal of Sports Medicine and Physical Fitness, 2026;66(4):523–531. DOI 10.23736/S0022-4707.25.17470-7, PMID 41661160
7. How Risky Is Mixed Martial Arts? Injury Rates and Patterns in Competitive Versus Recreational Athletes — Groessing L, et al., Healthcare (Basel), 2026;14(3):409. DOI 10.3390/healthcare14030409, PMID 41682259, PMCID PMC12898159
8. Incidence of Health Problems in Australian Mixed Martial Arts and Muay Thai Competitors: A 14-Month Study of 26 Combat Sports Events — Doherty CS, Barley OR, Fortington LV, Sports Medicine – Open, 2025;11(1):60. DOI 10.1186/s40798-025-00880-3, PMID 40437191
9. Discrepancies in combat sports injury reporting: A comparison of athlete self-reports and ringside physician data — Doherty CS, Barley OR, Fortington LV, Journal of Science and Medicine in Sport, 2026;29(3):222–230. DOI 10.1016/j.jsams.2025.08.018, PMID 40925822
10. IOC consensus statement: methods for recording and reporting of epidemiological data on injury and illness in sport 2020 (including STROBE Extension for Sport Injury and Illness Surveillance, STROBE-SIIS) — Bahr R, Clarsen B, Derman W, et al., Orthopaedic Journal of Sports Medicine, 2020;8(2):2325967120902908 (co-published British Journal of Sports Medicine 2020;54(7):372–389). DOI 10.1177/2325967120902908, PMID 32118084
11. Craniofacial and traumatic brain injuries in mixed martial arts — Fares MY, et al., The Physician and Sportsmedicine, 2021;49(4):420–428. DOI 10.1080/00913847.2020.1847623, PMID 33151104
12. Upper limb injuries in mixed martial arts — Fares MY, Baydoun H, Elhassan B, Abboud JA, The Physician and Sportsmedicine, 2023;51(5):434–441. DOI 10.1080/00913847.2022.2123257, PMID 36093854
13. Assessing the Incidence of Head Trauma in Australian Mixed Martial Arts: A Retrospective Analysis of Fight Outcomes — Brown DA, Gross G, Sports Health, 2025;17(4):689–696. DOI 10.1177/19417381241263332, PMID 39095970
14. Characteristics of Facial Trauma in Professional Mixed Martial Arts — Jones AJ, Hasnain F, Shipchandler TZ, Vernon DJ, Elghouche AN, Facial Plastic Surgery & Aesthetic Medicine, 2023;25(4):332–337. DOI 10.1089/fpsam.2022.0097, PMID 36413026
15. Eye Trauma in Mixed Martial Arts and Boxing — Chang CC, et al., Ophthalmic Plastic and Reconstructive Surgery, 2024;40(1):75–87. DOI 10.1097/IOP.0000000000002510, PMID 37656909
16. Characterising ocular injuries in competitive combat sports in Texas: a retrospective case-control study — Akanno UE, Malik M, et al., BMJ Open, 2025;15(10):e098197. DOI 10.1136/bmjopen-2024-098197, PMID 41125273
17. Injuries to Professional and Amateur Kickboxing Contestants: A 15-Year Retrospective Cohort Study — Lystad RP, Orthopaedic Journal of Sports Medicine, 2015;3(11). DOI 10.1177/2325967115612416, PMID 26740949
18. Brazilian Jiu Jitsu, Judo, and Mixed Martial Arts Injuries Presenting to United States Emergency Departments, 2008-2015 — Stephenson C, Rossheim ME, Journal of Primary Prevention, 2018;39(5):421–435. DOI 10.1007/s10935-018-0518-7, PMID 30043324
19. Sex differences in martial arts injuries presenting to emergency departments in the United States from 2015 to 2024 — Hamstra-Wright KL, Kremer H, Coumbe-Lilley JE, Khosla S, Injury, 2026;57(10):113658. DOI 10.1016/j.injury.2026.113658, PMID 42715919
20. Maxillofacial Trauma in the Ultimate Fighting Championship — Sifuentes-Cervantes JS, et al., Journal of Oral and Maxillofacial Surgery, 2021;79(9):1921.e1–e7. DOI 10.1016/j.joms.2021.05.009, PMID 34153246
21. Prevalence, Patterns, and Characteristics of Eye Injuries in Professional Mixed Martial Arts — Fliotsos MJ, et al., Clinical Ophthalmology, 2021;15:2759–2766. DOI 10.2147/OPTH.S319025, PMID 34211265
22. Craniofacial Injuries in Combat Sports: Patterns Across Boxing, Wrestling, and Martial Arts — Mehra S, Mangal R, Mirsky NA, Thaller SR, Journal of Craniofacial Surgery, 2026;37(3-4):468–473. DOI 10.1097/SCS.0000000000011751, PMID 40778941

Read next