Injury rates in BJJ and grappling, and why they differ from striking

1. The question behind the question

When a competitor asks how often people get hurt in jiu-jitsu, they are usually asking one of three different things and have not separated them: how likely am I to be hurt badly enough to stop training, how likely am I to be hurt badly enough to need surgery, and how likely am I to be hurt at all in the ordinary aches-and-tweaks sense. Those three questions have three different answers, and the published literature answers whichever one the authors chose to define — usually without flagging that the choice was the study's most consequential methodological decision.

The International Olympic Committee's 2020 consensus statement on recording and reporting injury and illness exists to force that decision into the open. It 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," and injury more narrowly as "tissue damage or other derangement of normal physical function due to participation in sports, resulting from rapid or repetitive transfer of kinetic energy." It then lays out three recording approaches — all-complaints, medical-attention and time-loss — and it is explicit that there is "no single approach to expressing risk appropriately for all sports injury surveillance projects."

That 2020 statement is the current general version. Checked on 22 September 2026, no 2025 or 2026 replacement has been published, and while sport-specific extensions exist for tennis and football, no grappling, judo or jiu-jitsu extension was found. That absence is not a footnote. It is the structural reason BJJ numbers cannot be pooled: there is no agreed template, so every research group builds its own, and the resulting figures are not comparable even when they appear side by side in the same table.

2. The definition moves the count by an order of magnitude

The judo review is the cleanest demonstration anyone has produced in a grappling sport, and it is worth stating precisely what it did. Mooren and colleagues gathered twenty-five studies of injuries during judo tournaments — nine rated good quality, ten fair, six poor — and sorted their findings by the definition each had used.

Under medical-attention definitions: 10.9 to 115 injuries per 1000 athlete-exposures, and 2.5% to 72.5% of athletes injured. Under time-loss definitions: 4.2 to 60 per 1000 athlete-exposures, and 1.1% to 4.1% of athletes injured. Competition only, mixed sex and age, IJF-type rulesets throughout.

Sixteen of the twenty-five studies used the medical-attention framing — "a physical complaint, for which assistance was sought from a tournament healthcare professional," or a close variant. That is the modal definition in grappling research, and the review names its weakness directly: "an inherent limitation of this definition is that it does not take injury severity into account." A taped finger and a torn anterior cruciate ligament are one count each.

The review's summary sentence is the one to carry into every other number in this article: "The reported injury incidence proportions in judo are largely dependent on the injury definition and methodological approach." Judo has been studied far more systematically than BJJ. If the definitional problem is this large in judo, it is larger in jiu-jitsu, where surveillance is weaker.

There is a second, quieter effect hiding inside the medical-attention approach. It measures not only injury but medical availability. A tournament with three physiotherapists and an orthopaedic team at matside will record more injuries than an identical tournament with one first-aider and a bag of ice, because more complaints have somewhere to go. Part of what these rates track is how well a competition was staffed.

3. Exposures, hours and matches are not the same unit

The denominator is the second half of the problem and it is treated even more carelessly than the definition.

The IOC consensus prefers time-based exposure where it can be had: "incidence-based measures that provide a standard time-window for the population at risk (injuries per hour) are preferable to measures for which the time at risk varies across individuals." An athlete-exposure — one athlete taking part in one session or one match — is a unit of wildly variable length. A first-round submission at forty seconds and a black-belt final that goes to overtime are one exposure each.

This matters because the same BJJ cohort produced both of the figures that get quoted against each other. In Stegerhoek's 881-participant global survey, using one three-part injury definition throughout, training incidence was 5.5 per 1000 training hours (95% CI 4.9–6.1) and competition incidence was 55.9 per 1000 matches (95% CI 38.8–73.0). It is tempting to read that as competition being ten times more dangerous. It is not a legitimate division. A match is minutes; an hour is an hour. The two rates share a cohort and a definition but not a unit, and dividing them produces a number that means nothing.

Where the training-versus-competition ratio has been measured with a consistent denominator, it is smaller and much more stable. NCAA men's wrestling surveillance from 1988-89 through 2003-04 recorded 26.4 injuries per 1000 athlete-exposures in matches against 5.7 in practice — a rate ratio of 4.6 (95% CI 4.4–4.8), across 9,723 injuries. The more recent 2014-15 to 2018-19 NCAA window puts the overall rate at 8.82 per 1000 athlete-exposures with a competition-versus-practice incidence rate ratio of 4.11 (95% CI 3.72–4.55). Both are men's folkstyle wrestling under time-loss-based reporting. Roughly four to five times, not ten, and not thirty.

4. What BJJ competition surveillance has actually measured

There are two serious on-site BJJ competition datasets in the published literature, and they disagree by a factor of nearly three because they were built differently.

The first is Scoggin and colleagues' coverage of eight statewide Hawaii tournaments between 2005 and 2011, with board-certified orthopaedic surgeons assessing at the venue. Across 2,511 matches and 5,022 exposures they recorded 46 injuries: 9.2 per 1000 exposures, under a definition of injuries requiring on-site medical care. The orthopaedic subset — 36 injuries, 78.3% of the total — worked out to 7.2 per 1000 exposures. Gi competition under a state-level IBJJF-style ruleset, mixed ages, no confidence interval reported.

That study also produced the only clean adolescent-versus-adult split in the BJJ literature: competitors aged fifteen and under sustained 4.1 injuries per 1000 exposures (5 injuries across 1,234 exposures), against 10.8 per 1000 exposures for those sixteen and over (41 across 3,788). Roughly two and a half times higher in adults. Five injuries is a fragile foundation and the finding should be held loosely.

The second is Kreiswirth and colleagues at the 2009 World Jiu-Jitsu No-Gi Championship: 951 male athletes aged 18 to 50, 1,606 athlete-exposures, 62 injuries of which 40 were joint injuries, giving a joint-injury rate of 24.9 per 1000 athlete-exposures. Male only by design, under the IBJJF no-gi ruleset as it stood in 2009 — which banned heel hooks at every belt. Knee and elbow tied as the most injured joints at 7.5 per 1000 athlete-exposures each.

That study is the source of the belt gradient people quote: blue 21.5, purple 21.3, brown 25.2, black 35.1 per 1000 athlete-exposures. It looks like a clean ascending line and it is not statistically significant. Brown and black against blue and purple gave an incidence rate ratio of 1.65 with a 95% confidence interval of 0.9 to 2.9 and P = .06. The interval crosses one. The gradient is a hypothesis, not a finding, and it should not be repeated as established.

These figures are abstract-level. The full text of the Kreiswirth paper could not be retrieved, so nothing here goes beyond what the abstract states.

5. Nearly all the exposure is in the training room

If a competitor only reads one section, this is the one.

Stegerhoek's global cross-sectional study of 881 BJJ practitioners recorded 1,913 injuries in total. Of those, 1,711 — about 89% — happened in training, not competition. Seventy-nine per cent of the training injuries occurred during free sparring; 57.5% were acute-onset and 63.4% involved direct contact. The definition throughout was medical attention, or at least one week of time loss, or at least two weeks of modified training, self-reported retrospectively.

A second large survey points the same way by a different route. Hinz and colleagues surveyed 1,140 BJJ athletes across 62 countries, 88.9% male, 63.9% regular competitors, using a stricter definition — injury causing at least two weeks away from training — over three years of recall. They recorded 1,052 injuries in 784 of the 1,140 athletes, which is 308 injuries per 1000 athletes per year, and summarised it as "2 out of 3 athletes reporting at least 1 injury within a 3-year period." Mechanisms: 77.6% during sparring, with submissions responsible for 29.7% and takedowns for 26.4%. Training and competition were combined in that count, and the figures are abstract-level only.

Two independent surveys, two different definitions, two different recall windows, and both land on the same structural fact: sparring is where the injuries are. Not the podium. Not the tournament. The Tuesday night round robin.

This is the finding that should drive how a competitor plans a preparation, and it is the one the sport talks about least. The volume and intensity of hard sparring a camp can carry is a training-design question, not a toughness question, and the injury data is pointing at it from every direction.

6. A worked scenario, and then the same scenario broken

Take a thirty-year-old blue belt who trains five hours a week, forty-six weeks a year — 230 training hours — and enters four tournaments a year at roughly three matches each, so twelve matches.

Apply Stegerhoek's two figures, which are the only pair in the BJJ literature drawn from one cohort under one definition:

• Training: 230 hours × 5.5 per 1000 = about 1.3 injuries per year
• Competition: 12 matches × 55.9 per 1000 = about 0.7 injuries per year

Roughly two-thirds of this athlete's expected injuries come from the room, not the bracket. That is the same shape wrestling shows from an entirely separate surveillance system, where competition is four to five times more dangerous per exposure but practice supplies far more of the exposures: the NCAA's 1988–2004 dataset contains 6,626 practice injuries against 3,097 match injuries.

Now break the scenario deliberately, because the breaking is the point.

Swap the competition figure for Scoggin's 9.2 per 1000 exposures — the on-site orthopaedic, requires-medical-care number from the Hawaii gi tournaments. Twelve matches × 9.2 per 1000 = 0.11 injuries per year. The same athlete, the same twelve matches, and the expected injury count has fallen roughly six-fold. Nothing changed but the injury definition and the medical staffing of the event.

Then put the judo numbers beside it: the same tournaments counted 10.9–115 per 1000 athlete-exposures under medical-attention and 4.2–60 under time-loss. Neither BJJ number is wrong. They are answers to different questions, and the reader now knows which question to ask before believing any grappling injury statistic — including every one printed in this article.

One caveat travels with the arithmetic and cannot be detached from it: it assumes constant risk per hour and per match, which no source establishes, and Stegerhoek's rates come from a sample that was 817 male out of 881, so a female reader should treat the output as unvalidated for her.

7. What actually gets injured

The body-region profile is the most stable finding in the whole grappling literature, and it is stable because it follows directly from mechanism.

In Stegerhoek's 888 detailed BJJ injuries: knee 223 (25%), shoulder 114 (13%), hand 71 (8%), chest 62 (7%). Training and competition combined, self-report, three-part definition.

In Hinz's survey under a two-week time-loss definition: lower extremity 45.7%, upper extremity 30.2%, knee alone 27.1%, with the most frequent knee injuries being meniscus (n=65), anterior cruciate ligament (n=36) and medial collateral ligament (n=36). The authors noted that ACL tears were associated with long return-to-sport times. Abstract-level.

In Scoggin's on-site tournament coverage, of 36 orthopaedic injuries: elbow 38.9% (14), knee 19.4% (7), foot/ankle 13.9% (5), hand 11.1% (4), shoulder 11.1% (4), with one hip and one cervical injury. Their summary sentence: "The elbow was found to be the joint most commonly injured during BJJ competitions." Thirty-six is a small number and those percentages are fragile — a single reclassified injury moves them by nearly three points. The non-orthopaedic remainder was mostly rib and costochondral injury (7 of 46, 15.2%) and lacerations (3, 6.5%).

Judo's competition profile differs, and differs in a way that tells you what judo is. The twenty-five-study review found head the most common injury location, then hand, knee, elbow and shoulder, with sprains most frequent, followed by contusions, lacerations, strains and fractures. More importantly, judo injuries occurred 50.0% to 84.9% in standing exchanges (tachi-waza) against 0.0% to 33.3% on the ground (ne-waza), and every included article agreed that standing dominated. An older systematic review of judo put average injury risk at the 2008 and 2012 Olympic Games at roughly 11–12% of athletes, with being thrown the most common mechanism, and severe injuries described as "quite rare" but usually affecting brain and spine.

Wrestling lands somewhere between. NCAA men's wrestling in the 2014-15 to 2018-19 window put knee at 21.4%, shoulder at 13.4% and head/face at 13.3%, with concussion the single most prevalent specific injury — a reminder that folkstyle wrestling is not a joint-only sport. The historical 1988–2004 dataset found match injuries running over 40% lower extremity, 26% upper extremity and 17% head and neck, with the largest single share of all injuries attributed to "direct contact during a takedown (42.3%)."

That is the through-line. Where the sport spends its time standing and throwing, the injuries move toward heads, necks and impact. Where it spends its time on the ground attacking joints, the injuries move toward knees, elbows and shoulders.

8. Submissions, and the arm bar in particular

Submission holds accounted for 28% of injuries (n=247) in Stegerhoek's cohort, with armlocks and leglocks at 37% of those each (n=92 apiece). Broken down by specific hold, the armbar accounted for n=52 (21% of submission injuries), the toe hold n=24 (10%) and the inside heel hook n=20 (8%).

Do the arithmetic that those numbers invite. Submissions were 28% of all injuries, and the inside heel hook was 8% of submission injuries — so the inside heel hook is roughly 2% of all injuries in a cohort of 881 self- reporting practitioners, gi and no-gi pooled, under a three-part injury definition. That is worth holding onto before reading the next section, because the discourse around leg attacks is loud and the measured share is small.

The arm bar's dominance shows up independently in the on-site tournament data. Scoggin's group attributed 10 of 14 elbow injuries to the arm bar, with the mechanism being elbow hyperextension across the opponent's pelvis. That is the single most reproducible injury mechanism in competitive BJJ and it is not a leg lock. Nothing in this article describes how any of these holds are applied, entered, defended or escaped; the mechanism is stated because it is what the surveillance recorded.

9. The leg-lock rule change, and what the numbers can carry

On 1 January 2021 the IBJJF made heel hooks and knee reaping legal in adult brown and black belt no-gi divisions. That created something rare in sports medicine: a rule change with a before and an after.

Piekarski and colleagues used it. Their cohort was brown and black belt IBJJF competitors — 303 adult competitors (defined by the study as under thirty years old) and 455 masters competitors (thirty and over) from 2021, and 164 adult brown and black belt competitors from 2009 as a comparison arm. The denominator was matches.

Their headline result: knee injury ran 26.5 per 1000 matches in exposed competitors against 2.2 per 1000 matches in unexposed competitors, relative risk 12.0 (95% CI 1.5–96.1; P < 0.001).

That confidence interval is the most important thing on this page. It runs from 1.5 to 96.1 — a factor of sixty-four. The study could not distinguish "modestly more risky" from "catastrophically more risky." The honest reading is that exposure to heel hooks was associated with knee injury risk somewhere between modestly and enormously higher, and the phrase "twelve times" should never appear without the interval welded to it, because printed alone it makes the most over-claimed statistic in modern grappling look precise.

Two further results from the same abstract complicate the picture in both directions. Ankle injury ran 19.8 per 1000 matches in exposed against 8.8 in unexposed, with no significant difference. And there was no significant difference in knee or ankle injury incidence between the 2009 and 2021 athletes overall — the rule change appears to have raised risk for those exposed to the technique without the study detecting a population-level shift. Both halves belong in any honest summary. The authors' own framing of the evidence base is worth quoting: "While evidence suggests that the heel hook can result in severe ligamentous damage to the knee and/or ankle, definitive data on injuries from this technique are lacking."

All of that is abstract-level. The full text was not retrievable from either the publisher or PubMed Central, so the abstract's description of the cohort is the limit of what can be said: brown and black belt IBJJF competitors, 2021 against 2009. The abstract names no championship, never says "no-gi," and describes no ascertainment method. Secondary write-ups that identify the specific events or the clinical method are going beyond the source. A separate article covers the leg-lock question in depth, including how the knee injuries themselves present and what the rule sets across federations actually permit.

10. What the IBJJF rulebook says, as of today

Rules are not epidemiology, but they are the only lever anyone has actually pulled, so it is worth stating their status precisely rather than from memory.

The current published edition of the IBJJF Rules Book carries a cover date of JUN.2024 and a footer reading "VERSION 6.1 2024". That PDF was downloaded from the federation's own books-and-videos page on 22 September 2026; no 2025 or 2026 edition exists. Note the small discrepancy that trips people up: the site's link text reads "v6.0" while the document's own footer reads "VERSION 6.1 2024." The document is the authority.

Within it, the Technical Fouls — Illegal Moves table is laid out in six division columns: "4 to 12 years old"; "13 to 15 years old"; "16 & 17 years old (all ranks) and white belts (Adult to Master 7)"; "Adult to Master 7 (blue & purple belts)"; "Adult to Master 7 (brown & black belts) except Adult No Gi"; and "Adult (brown & black belts) No Gi". Heel hook, locks that twist the knee, knee reaping and the toe hold with outward pressure all sit in that table. The existence of a dedicated "Adult (brown & black belts) No Gi" column alongside a column explicitly excluding that same group is the structural proof of the carve-out.

The federation's own announcement dates it: "Heel Hooks and Knee Reaping for Black and Brown Belt adult NoGi Divisions," valid starting 1 January 2021. Adopted and in force from that date, and still reflected in the current v6.1 rulebook as verified on 22 September 2026.

Two things about how that rulebook is written are worth a competitor's attention. First, knee reaping is banned positionally rather than by technique name — the rulebook defines it by a specific geometry of thigh, calf, foot and trapped leg, which means the prohibition catches configurations regardless of what anyone calls them. Second, the same illegal-moves table carries slams and spinal locks without a choke, and the rulebook separately makes grounds for disqualification of "the suplex movements that will project or force the opponent's head or neck into the ground." The federation's rule-writing treats head-and-neck impact as categorically different from joint attacks, which is consistent with what the judo mechanism data shows.

11. What real surveillance looks like, and why BJJ does not have it

Wrestling has the National Collegiate Athletic Association's injury surveillance programme: athletic trainers reporting into a common system, across seasons, with a stable definition and a counted denominator. Judo has two systematic reviews pooling dozens of tournament studies. Brazilian jiu-jitsu has surveys and single-tournament snapshots.

That asymmetry is itself the finding. There is no prospective training surveillance in BJJ at all. Every BJJ training figure in this article is recall-based — someone remembering, months or years later, what happened to them. Recall-based injury data tends to lose the minor and the ambiguous and keep the dramatic, and there is no way to know from the outside how much of each it lost.

It compounds. Because there is no grappling-specific extension to the IOC surveillance consensus, no two research groups are obliged to define injury the same way, which is why the BJJ competition rates in section 4 differ by a factor of nearly three. The fix is structural and it is not the reader's to make, but knowing the fix is missing is what stops a number being over-trusted.

The two large surveys even disagree with each other about belt rank. Stegerhoek found injury incidence significantly lower for brown and black belts than for white and blue belts (p=0.001). Hinz found that higher belt rank was a significant risk factor (P=.003), alongside competing regularly (P=.003) and older age (P<.001). Both are self-report surveys, both are large, and they point in opposite directions. The correct thing to print is the disagreement, not one side of it. Kreiswirth's competition belt gradient, which leans toward the Hinz direction, was not statistically significant.

One footnote for anyone checking sources: PubMed's MEDLINE abstract for the Stegerhoek paper renders the belt sentence as a higher rate among higher belts, which is the opposite of what the publisher's abstract and the paper's Results section say. The BMJ Open Sport & Exercise Medicine version is the one to read.

12. Skin, throat and the categories that fall through

Two categories of grappling harm are badly served by injury surveillance, for opposite reasons.

Skin infection is surveilled well — in wrestling. NCAA men's wrestling across seventeen programmes and thirty-five team-seasons from 2009-10 through 2013-14 recorded 112 infections in 87 athletes across 78,720 athlete-exposures: 14.23 per 10,000 athlete-exposures (95% CI 11.59–16.86), athletic-trainer-reported viral, bacterial and fungal infections. Of those, 89.3% were identified during practice, 67.9% occurred in the regular season, 74.1% caused at least twenty-four hours of time loss, and 22.3% were recurrent. Viral infections ran 1.72 times the bacterial rate (95% CI 1.09–2.72) and 2.08 times the fungal rate (95% CI 1.28–3.39), while fungal infections more often cost under a day (75.0% versus 12.5%; proportion ratio 6.00, 95% CI 3.30–10.92).

The most useful part of that dataset is not the average. 65.2% of the infections came from five team-seasons, each contributing between eleven and nineteen. Infection risk clustered by room, not evenly across the sport. The authors' own conclusion is that the findings "highlight the contagiousness of skin infections and suggest that skin infection rates may be attributable to high incidences among particular teams." The older NCAA dataset found skin infection was "the most commonly reported time-loss condition, accounting for more than 17% of reported events" in wrestling practices, primarily herpes simplex and ringworm.

Those are wrestling numbers. They are cited here as wrestling numbers and they do not transfer. No BJJ skin-infection surveillance with a denominator was found, and anyone quoting a BJJ infection rate is quoting something that has not been measured.

Throat and larynx injury is the opposite case: almost certainly common, and essentially unmeasured. A mixed-methods study recruited 160 BJJ and MMA practitioners through social media over six weeks and found 88% reporting laryngopharyngeal symptoms after chokeholds — sore throat 79%, painful swallowing 66%. Sixty-eight per cent resolved within a week; 15% reported a permanent voice change; 87% trained while symptomatic; and only 11% (n=18) sought care, of whom nearly half (n=8) were diagnosed with a hyolaryngeal fracture.

Every one of those percentages is a prevalence among self-selected respondents to a social-media recruitment, with no exposure denominator at all. Selection bias is severe: people with throat symptoms are far more likely to answer a survey about throat symptoms. These are not rates and they are not risk estimates. What the study does establish, and what the authors say, is that "laryngopharyngeal symptoms are common among athletes yet often disregarded, with athletes frequently continuing to train despite symptoms and seldom seeking care."

That last clause is the part worth acting on. Throat symptoms after chokes are common and rarely taken to anyone. Hoarseness, swallowing pain or a voice change that does not settle is a reason to see a clinician — not a reason to self-assess, and not something this article can help anyone evaluate. Nothing here describes how a choke produces those injuries, and nothing here is guidance about when to tap.

The emergency-department picture has the same denominator problem. A survey-weighted analysis of US emergency departments from 2014 to 2023 produced a national estimate of 36,635 presentations for head, neck and spine injuries from martial arts, mean age 24.5 years, with jiu-jitsu associated with 26.9% and karate with 26.8% — the two most frequently named styles. Lumbar strain (19.8%) and cervical strain (18.3%) were the most common injuries, concussions were 18.4%, and throwing or flipping an opponent was the predominant mechanism at 17.1%. That describes the mix of who arrives at an emergency department. With no participation denominator it cannot produce a rate, and it cannot rank martial arts by danger — only by how many of their participants show up.

13. Is BJJ safer than MMA? The comparison, honestly

The meta-analytic pooled competition injury rate for mixed martial arts is 228.7 per 1000 athlete-exposures (95% CI 110.4–473.5), from six studies. The authors of that meta-analysis flag their own definitional problem in plain language: in all but one of the included studies, "what constituted a reportable injury was at the discretion of the attending ringside physician."

Set that against BJJ competition rates of 9.2 or 24.9 or 55.9 per 1000, each under a different definition and a different denominator, and the twenty-six-fold ratio people quote writes itself. It should not. The BJJ figure in that comparison is on-site medical care at Hawaii gi tournaments; the MMA figure is ringside-physician discretion at professional fights. Different definition, different denominator population, different ruleset, different medical infrastructure. The ratio is an artefact of the methods, not a measurement of the sports.

What the MMA data does establish is the profile. Head injuries were 66.8% to 78.0% of MMA competition injuries, wrist and hand 6.0% to 12.0%, with laceration and abrasion at 36.7% to 59.4%, fracture 7.4% to 43.3% and concussion 3.8% to 20.4%. The same review offers comparator figures — judo 44.0, taekwondo 79.4, amateur boxing 77.7, professional boxing 118.0 to 250.6 per 1000 athlete-exposures — and those are all cross-study with mismatched definitions. They are usable only to say that the pattern differs, never to rank the rates.

The authors' own framing is the defensible claim: the MMA injury pattern is "very similar to that in professional boxing" but "unlike that found in other combat sports such as judo and taekwondo," because head injuries "are relatively uncommon in judo and taekwondo where punches to the head are disallowed."

So the honest answer to the question everybody types is this. Grappling and striking injure different things. A grappler's expected injuries are knees, elbows and shoulders, and the most likely single mechanism is a joint taken past its range or a takedown landing badly. A striker's expected injuries are the head and the face: lacerations, fractures, concussion. That is a real and mechanistically obvious difference, and it is a sound basis for deciding which risks you are willing to carry. What it is not is a rate comparison, because no two of these studies share an injury definition or a denominator. Anyone quoting "N times safer" is quoting the methods section of two unrelated papers and calling it a result. The consequences of striking-specific damage are a different category of problem with a different literature behind them.

14. Who is missing from all of this

Every rate in this article comes from a narrower population than the sentence containing it implies, and three groups are missing almost entirely.

Women. Stegerhoek's cohort was 817 male of 881, leaving 64 women — 7.3%. The paper reports training incidence per 1000 hours as similar between sexes, but that comparison rests on those 64 people. Hinz's was 88.9% male. Kreiswirth's was male only, by design. Every NCAA wrestling figure quoted here is men's wrestling; no women's collegiate wrestling surveillance figures were retrieved, despite the sport's rapid growth. The judo review is the only source with a real sex comparison, and it is internally inconsistent: six studies found men injured more, six found women more, and three reported more knee sprains, including anterior cruciate and medial collateral ligament injuries, in female athletes. There is no adequately powered BJJ injury study in women, and no coefficient exists that would let anyone scale the male numbers across.

Adolescents. The cleanest BJJ figure is Scoggin's 4.1 per 1000 exposures in competitors fifteen and under, built on five injuries. The judo review reports adult incidence proportions of 1.3%–21.0% against youth 0.9%–14.4% across four studies, three of which favoured adults. No prospective BJJ youth surveillance exists. The IBJJF's rulebook runs four separate youth columns in its illegal-moves table, which means the federation treats youth risk as categorically different — but no data anchors that treatment.

Hobbyists, who are the overwhelming majority of the sport. Every competition rate here comes from tournament entrants. Both large surveys sampled people engaged enough with jiu-jitsu to fill in a research questionnaire. The person who trains twice a week, never enters anything and goes moderately in sparring is measured by no study on this list. Stegerhoek's 5.5 per 1000 training hours is the closest available proxy, and it is retrospective self-report from a competition-adjacent sample.

One further gap is worth naming because it sits at the intersection of two things grapplers do at once. An older judo review noted that some studies "suggested a relationship between nutrition, hydration and/or weight cycling and judo injuries," and that is as far as the retrieved evidence goes. No study quantifying injury risk in a dehydrated or acutely weight-reduced state in any grappling sport was found — which matters, because the BJJ weigh-in format puts athletes on the mat minutes after the scale. Nothing here establishes that this raises injury risk. Nothing here establishes that it does not.

And finally: cauliflower ear, the injury most visibly associated with grappling, has no incidence figure with a denominator in any grappling sport in the retrieved literature. An older judo review names auricular injury as a typical chronic complaint alongside finger joints and the lower back. That is all there is.

What we could not verify

Several widely circulated figures did not survive being chased to their source, and a few are worth naming so they can be recognised in the wild.

"BJJ injury incidence is 26 times lower than MMA." Traced to the discussion section of the Hawaii tournament paper, which compares its own on-site-medical-care-at-a-gi-tournament rate of 9.2 per 1000 exposures against MMA rates built on ringside-physician discretion. Different definition, different denominator population, different ruleset. The ratio is not a measurement and is not printed here in any form.

"Heel hooks had an injury rate of 36.4 per 1000 matches in adults" and the associated 14.5 per 1000 combined figure. These appear in search-engine summaries attributed to the Piekarski study. They are not present in the fetched abstract, and the full text is behind a 403 at both the publisher and PubMed Central. The verified figures from that abstract are the ones in section 9 and nothing more.

"Heel hooks increase knee injury risk 12-fold." The point estimate is real; presenting it as a finding is not, because the 95% confidence interval runs from 1.5 to 96.1. Printed alone it misrepresents a study that could not tell 1.5× from 96×. This article prints the interval every time the estimate appears.

"Judo competitions report 25.2 to 130.6 injuries per 1000 exposures." Surfaced in a search summary. The ranges actually verified in the judo systematic review are 10.9–115 under medical-attention and 4.2–60 under time-loss definitions. The origin of 130.6 could not be located and it is not printed.

"BJJ is safer than judo because there are no throws." No source supports this. Judo's injuries do concentrate in standing exchanges, and BJJ takedowns caused 26.4% of injuries in the Hinz survey. BJJ has throws; it does fewer of them. The mechanism observation is printable, the safety conclusion is not.

Any BJJ skin-infection rate. None exists with a denominator. The wrestling figures above are cited as wrestling and were not transferred.

Any belt-level risk claim beyond the reported disagreement. Two large primary surveys point in opposite directions and the one competition gradient was not statistically significant.

Emergency-department counts as injury rates. "Jiu-jitsu was associated with 26.9% of martial-arts head, neck and spine emergency presentations" is printable. "Jiu-jitsu is the most dangerous martial art for the spine" is not, because the sampling frame has no participation denominator.

Beyond the individual claims, two structural gaps shape everything above. There is no prospective training surveillance in Brazilian jiu-jitsu at all — not one study — so every BJJ training figure here is recall-based. And there is no grappling-specific extension to the IOC injury-surveillance consensus as of 22 September 2026, which is the reason these numbers cannot be pooled. Eight of the sources behind this article were readable only at abstract level, and nothing in this article states body-text detail that an abstract cannot support.

Sources

Sourced to

1. International Olympic Committee 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) — Bahr R, Clarsen B, Derman W, et al., British Journal of Sports Medicine, 2020;54(7):372–389. DOI 10.1136/bjsports-2019-101969, PMID 32071062
2. Epidemiology of Injuries during Judo Tournaments — Mooren J, von Gerhardt AL, Hendriks ITJ, Tol JL, Koëter S, Translational Sports Medicine, 2023;2023:2713614. DOI 10.1155/2023/2713614, PMID 38654918
3. Assessment of Injuries During Brazilian Jiu-Jitsu Competition — Scoggin JF 3rd, Brusovanik G, Izuka BH, Zandee van Rilland E, Geling O, Tokumura S, Orthopaedic Journal of Sports Medicine, 2014;2(2):2325967114522184. DOI 10.1177/2325967114522184, PMID 26535299
4. Incidence of injury among male Brazilian jiujitsu fighters at the World Jiu-Jitsu No-Gi Championship 2009 — Kreiswirth EM, Myer GD, Rauh MJ, Journal of Athletic Training, 2014;49(1):89–94. DOI 10.4085/1062-6050-49.1.11, PMID 24377959
5. Injury prevalence among Brazilian Jiu-Jitsu practitioners globally: a cross-sectional study in 881 participants — Stegerhoek PM, Brajovic B, Kuijer PPFM, Mehrab M, BMJ Open Sport & Exercise Medicine, 2025;11(1):e002322. DOI 10.1136/bmjsem-2024-002322, PMID 40092168
6. Injury Patterns, Risk Factors, and Return to Sport in Brazilian Jiu Jitsu: A Cross-sectional Survey of 1140 Athletes — Hinz M, Kleim BD, Berthold DP, et al., Orthopaedic Journal of Sports Medicine, 2021;9(12):23259671211062568. DOI 10.1177/23259671211062568, PMID 34988235
7. Knee Injury in Competitive Brazilian Jiu Jitsu Athletes: Implications for Training — Piekarski M, Kreiswirth E, Barber Foss K, Jimenez ML, Myer GD, Montalvo AM, Sports Health, 2026. DOI 10.1177/19417381251400303, PMID 41549501
8. IBJJF Rules Book — General Competition Guidelines and Competition Format Manual (cover JUN.2024, footer "VERSION 6.1 2024") — International Brazilian Jiu-Jitsu Federation, June 2024 edition, PDF downloaded and verified 22 September 2026
9. New Rules Updates — "Heel Hooks and Knee Reaping for Black and Brown Belt adult NoGi Divisions," valid from January 1st, 2021 — International Brazilian Jiu-Jitsu Federation, retrieved 22 September 2026
10. Epidemiology of Injuries in National Collegiate Athletic Association Men's Wrestling: 2014-2015 Through 2018-2019 — Powell JR, Boltz AJ, Robison HJ, Morris SN, Collins CL, Chandran A, Journal of Athletic Training, 2021;56(7):727–733. DOI 10.4085/1062-6050-429-20, PMID 34280284
11. Descriptive Epidemiology of Collegiate Men's Wrestling Injuries: National Collegiate Athletic Association Injury Surveillance System, 1988-1989 Through 2003-2004 — Agel J, Ransone J, Dick R, Oppliger R, Marshall SW, Journal of Athletic Training, 2007;42(2):303–310. PMID 17710180
12. Epidemiology of Skin Infections in Men's Wrestling: Analysis of 2009-2010 Through 2013-2014 National Collegiate Athletic Association Surveillance Data — Herzog MM, Fraser MA, Register-Mihalik JK, Kerr ZY, Journal of Athletic Training, 2017;52(5):457–463. DOI 10.4085/1062-6050-52.2.16, PMID 28362160
13. 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, PMID 26535267
14. Injuries in judo: a systematic literature review including suggestions for prevention — Pocecco E, Ruedl G, Stankovic N, et al., British Journal of Sports Medicine, 2013;47(18):1139–1143. DOI 10.1136/bjsports-2013-092886, PMID 24255909
15. Fighting for safety: head and spine injuries in martial arts presenting to U.S. emergency departments — Rabon W, Tirmizi Z, Carnovale B, et al., Journal of Clinical Neuroscience, 2026. DOI 10.1016/j.jocn.2026.112306, PMID 42759081
16. Laryngopharyngeal Trauma in Grappling Martial Arts: A Mixed-Methods Study — Singerman KW, Bird CL, Cho S, Kavookjian HL, Kraft SM, OTO Open, 2026. DOI 10.1002/oto2.70256, PMID 42256725
17. Common grappling submissions: A descriptive, illustrative and literature review of anatomic structures at risk and pathophysiology — Hasegawa ME, Rimm JB, Ishikawa KM, et al., Hawai'i Journal of Health & Social Welfare, 2026. DOI 10.62547/rnvf9981, PMID 42245247

Read next