Chokes and safety

I have been teaching martial arts since 1992. I spent twenty-three years in the Australian Army and retired as a Major. I have been a senior instructor on Army unarmed combat courses, and I have been a Brazilian jiu-jitsu black belt under John Will since 2017. Of everything we do at Rogue, neck attacks are the thing I have thought about longest and the thing I am strictest about.

This page exists because people ask. Usually it is a partner, or a parent, and the question underneath is always the same one: is somebody going to choke my kid, and is that safe? You deserve a straight answer rather than a brochure. Here it is.

The one rule

Tap at pressure, not at pain.

If you have waited until it hurts, you have waited too long. That is not a figure of speech and it is not about toughness. It is about how the body actually signals damage.

Why pain is the wrong signal

Pain does not arrive when a joint is in danger. It arrives once damage has started.

• The elbow does not hurt until there is enough pressure on the capsule to tear something.

• A shoulder lock does not hurt until the tendons are already stretched.

• Many leg locks do not hurt until the leg is broken, and then they hurt.

So the skill is not tolerating pain. The skill is recognising the mechanics before the pain arrives, and that is something you can be taught.

There is an asymmetry here that catches people out. Discomfort shouts because it has weight and pressure behind it. Danger whispers, because there is nothing to feel until the ligament goes. The signal that is easiest to notice is the one that matters least.

If you rely on pain, you end up with loose joints very quickly. I had a friend who believed that taking a certain amount of pain in the shoulder was strengthening it: that not tapping straight away to arm locks would let him take more pressure over time. All it did was weaken the ligaments and tendons. After about three months, he could sneeze, and his shoulder would dislocate. He gave up jiu-jitsu.

Discomfort and danger are different things

Discomfort is significant pressure. It is unpleasant. Nothing is going to break.

Danger is a position where you can actually be hurt.

Learning the difference is most of the job, and people go wrong in both directions; some tap at every unpleasant sensation and never learn what they can carry, and some sit in a genuinely dangerous position because it does not hurt yet.

What to feel instead: the six core submissions

We teach six submission families, and each has a mechanical tell that arrives before any pain does. This is the reason the six exist: so you can recognise the mechanism rather than wait for a signal that comes too late.

For the Americana and the Kimura: when you get to pain, you are late.

For the three chokes, correctly applied, there should be no pain at all; only that building pressure. That pressure is the signal, and it is the one to tap to.

What a strangle actually does

A strangle is pressure applied to the sides of the neck. It reduces the blood supply to the brain. Consciousness needs that supply to be continuous; interrupt it enough, and the person loses consciousness and starts again a few seconds after the pressure comes off.

The mechanism isn't as simple as squeezing the carotid arteries, as is usually taught. Stellpflug and colleagues described this in 2020. Compressing the carotids lowers the pressure driving blood into the brain. The jugular veins on the way out are compressed by far less force than the arteries need, and blocking them raises the pressure inside the skull, which reduces the flow through the brain from the other direction. In some people, the carotid sinus reflex adds to it. Pressure on a sensitive spot in the artery wall triggers a drop in heart rate and blood pressure. That last one is uncommon and unpredictable, and it’s why I care about somebody's medical history. In study conditions, loss of consciousness has been reported within roughly six to fourteen seconds.

Two things follow from that.

The first is that it is fast. There is no long window in which you notice things are going wrong and decide to do something about it. By the time it is uncomfortable enough to worry you, you are most of the way there. That is the physiology behind the rule at the top of this page.

The second is that a strangle is a circulation technique, not an air technique. It should never be on the windpipe.

What it does not do

It does not crush the throat. Or it should not.

A well-applied strangle sits on the soft sides of the neck. If your partner is coughing, gagging, or hoarse afterwards, the technique was on the airway, and it was wrong; not "close enough", wrong. That distinction is the whole difference between a controlled skill and an injury, and the grappling world is worse at it than it likes to think. (Author's note: there are some strangles that deliberately apply pressure through the trachea, pushing it rearwards to reach the carotid arteries. These are inherently more dangerous and are generally taught only to more experienced practitioners. They will often elicit coughing and gagging. The problem is that these techniques have normalised incorrect applications of strangles in general.)

Singerman and colleagues surveyed 160 jiu-jitsu and MMA athletes in 2026 about throat and voice injuries. Eighty-eight per cent reported laryngopharyngeal symptoms after being caught in chokeholds; most often a sore throat (79 per cent) and pain on swallowing (66 per cent). Fifteen per cent reported a permanent change to their voice. Eighty-seven per cent kept training while symptomatic, and only eleven per cent sought medical care.

That is a survey: self-reported, self-selected, taken at one point in time, and it cannot tell you how common these injuries are across the sport as a whole. But it is not measuring a rare catastrophe. It is measuring an ordinary week in a lot of gyms, and what it describes is sloppy technique landing on the airway plus a culture that shrugs at the consequences. I am not interested in running that room.

Some variations do sit closer to the line than others. Certain D'Arce and anaconda entries when applied incorrectly can put pressure on the side or back of the neck, and front chokes often put pressure on the trachea when applied with the standard arm position. That is exactly why we teach the variations that build pressure in the head rather than the ones that produce a pain response, and why we apply them softly. We are working on soft tissue.

The other thing worth clearing up, the twitching people sometimes see after somebody has been put out is not the body pumping blood back to the heart. It is myoclonus; brief, irregular, multifocal muscle jerks that happen when the brain is briefly short of blood. It is a brainstem-level response, the same thing that happens in an ordinary faint, and it is not an epileptic seizure. It stops on its own within seconds. It looks far worse than it is, which is exactly why the people in the room need to know what they are looking at before they see it.

Why I treat this seriously

Because the injuries that do occur are not the ones people expect.

Hasegawa and colleagues published a review in 2026 of the anatomy at risk in common grappling submissions. For neck attacks the list is longer than most coaches would guess: the cervical spine, the carotid arteries and jugular veins, the nerves of the neck, the trachea and thyroid cartilage, and, when the arm is high, the jaw, the teeth and the temporomandibular joint. They note that the carotid can be injured by direct compression, by traction, or by twisting force, and that carotid and vertebral artery dissections, narrowing and ischaemic strokes have all been reported after chokes. They also note that the incidence is essentially unknown: what exists in the literature is case reports and small series, not rates. Their own practical conclusion is that the forces involved warrant control and caution, and that if you are caught, you should submit early.

Rowe and Wedlake, writing in 2009 for an audience of martial arts instructors, are worth reading in full if you teach. Their central point is that the dangerous events are the unlikely ones you cannot screen for by looking at somebody. An artery in the neck can tear and the stroke can arrive hours later, after training has finished and everyone has gone home. Arterial dissection is a leading cause of stroke in people under forty. Existing plaque in an artery wall can be disturbed by moderate force. The risk climbs with age and with how often you do it. Genetic connective tissue conditions raise it substantially in people who are otherwise young and fit. Their recommendations to instructors are unremarkable and correct: take a medical history, pair people carefully, teach the tap before anything else, stay off the upper neck, and know the signs of a stroke.

None of that makes strangles unusually dangerous compared with the rest of what happens on a mat. It makes them a technique whose failure mode is serious enough that you do not get to be casual about it. That is the whole argument.

What the evidence shows, and what it does not

There are versions of this article online that use the recent research to reassure you, and even infer that grappling may impart a health benefit. I am not doing that, because the research does not support it, and because you should be suspicious of anybody whose reading of the science happens to conclude that their product is good for you.

Here is the state of it, honestly.

So: two small studies pointing towards reassurance, one survey pointing at a lot of unreported throat injury, and a review saying the serious vascular events are real but uncounted. Nobody has followed a group of grapplers across decades of training. None of this work was done in Australia, or at a club like this one, or on anybody I have met.

I want to be blunt about the Stacey paper in particular, because it is the one that gets quoted at me. It is a genuinely interesting result and it is being used badly. A snapshot comparison of eleven elite athletes cannot show that jiu-jitsu caused their cerebral blood flow to be what it is; those athletes may simply have been different people before they ever started. The authors themselves offer years of high-intensity training as an equally likely explanation. And even taken at its most generous, it says nothing whatsoever about whether being strangled is safe.

Rogue does not claim that being choked is good for you. The honest landing point is that the science here is early and mixed. That is not a disappointing answer. It is the reason for every rule in the next section. If the evidence were settled and benign, care would be a courtesy. While it is unsettled, care is the only defensible position, and our rules do not depend on how the research settles. Never to unconsciousness, release on the tap, no cranking, nothing on the windpipe: those are correct whichever way it lands.

On police neck restraints

You may have read that police were banned from using these. That is true in Australia, everywhere. Queensland Police discontinued the Lateral Vascular Neck Restraint in April 2023, which made Queensland the last Australian jurisdiction to do so; the decision followed two deaths in custody over five years and a coroner's recommendation after an inquest.

Earlier versions of this page tried to have it both ways: one draft said police had banned it, another said police and military worldwide still teach it. I am not going to straddle that. In this country, police do not use it.

I am also not going to pretend it counts in my favour. The situations are genuinely different. A police officer applies force to a frightened, resisting, sometimes intoxicated stranger, usually alone, often having practiced the technique a handful of times in their whole career, with no tap available and no way for the other person to consent or to stop it. A training partner is consenting, sober, in a controlled room, with a coach watching, and can end it instantly. Those are not the same problem, and the ban does not transfer straight across to a mat.

But there is a lesson in it, and it runs the other way from the one people expect. Trained adults, under stress, got this wrong often enough and badly enough that every state and territory withdrew it. If that can happen to police officers, it can certainly happen to somebody in their sixth week of jiu-jitsu who is tired, competitive and gripping too hard. That is why the rules below are not suggestions, and why I do not negotiate them.

The contract

Everything here rests on an agreement that runs both ways, and it should be said out loud rather than assumed.

I need to know you will tap. Without that, submissions cannot be practiced at all. If every submission I applied broke something, I would feel terrible, and I would run out of training partners inside a month.

And when I tap, I need to know you will stop. "It's not on yet, so I'm going to keep going" is the worst thing that happens on a mat, because it breaks the trust in both directions at once.

The sentence underneath all of it, which every new student hears:

The only reason it isn't breaking is because your partner is choosing not to break it.

That is what a submission is. Not a win. A demonstration, held at the point where it could become an injury, by somebody who has decided not to take it there.

Why people don't tap

Three reasons, and only one of them is ego.

1. They don't recognise the position at all. The attack is unfamiliar enough that they never identify it as an attack. They do not see this is a Kimura.

2. They recognise it but feel no danger, because there is no pain yet. See the top of this page.

3. Ego. They do not want to lose, so they keep going.

Nobody should lose a joint through ignorance, and nobody should lose one through stubbornness either. So partners here are trained not to assume the tap is coming. That is the single most important habit in the room, and it is what makes the other two reasons survivable.

How we apply a submission

The protocol, and it is deliberate.

1. Rotate into the submission, then hold. Do not keep rotating.

2. Give them a second. They may wiggle.

3. Tell them. "I think you're done."

4. They tap, and usually apologise for leaving it late.

I would rather they were embarrassed than hurt. Ego gets in the way sometimes and people are late. That should cost them a moment of awkwardness rather than three months off the mat.

Control before submission

We prioritise control. The submission is secondary, and that is a safety decision as much as a technical one.

A submission thrown without control gives your partner no warning. We call it a Hail Mary: you are hoping, and they have no time to recognise danger before it arrives. Hunting submissions that way is frowned on here, and I will say so out loud.

Done properly it is the opposite: they are immobilised first, then the pressure comes onto the limb, so they register the threat and can answer it. The warning is built into the technique. A submission that gives your partner no chance to tap is not a good submission, however clean it looks, and it is not good self-defence either, because it only works on somebody who does not see it coming.

I stopped refereeing over this. I was officiating two blue belts, sixteen or seventeen years old, and one of them was about to have his knee completely exploded. I called the match and awarded the fight. The competitors protested. The senior referee confirmed the call had been within my rights; and then told me that next time I should just let him break the leg. That was the last time I ever refereed. I was not comfortable standing there watching people hurt each other. Elite athletes understand what they are choosing. These were kids, and one of them was about to end up in a brace needing knee surgery.

And the honest cost of training this way: from a pure competition standpoint it is probably poor preparation, because it removes surprise. I take that trade deliberately.

The rules at Rogue

1. Tap early. Tap on the position, not on the pain. If you have waited until it hurts, you have waited too long. Tap loudly on your partner, on the mat, or on yourself. If your hands are trapped, say the word, "tap", or tap with a foot. There is no cost to tapping here. Nobody keeps score. If you are not sure what is happening, tap and then ask.

2. Release instantly. The moment the tap comes, you let go. Not after you finish the grip. Not one more second. Instantly. This is the single thing I will stop a class for.

3. Nothing on the windpipe. Pressure goes on the sides of the neck. If your partner coughs, gags or sounds hoarse, it was wrong. Stop, and tell me, before you go back to it, and before you go home.

4. No cranking. Straight, slow, progressive pressure. No twisting the neck, no bending it fast, no jerking, nothing snapped on. A crank rotates the neck and loads the cervical spine; done hard enough it can break it. It is not a choke and it is not used in rolling here. A strangle should arrive gradually enough that your partner has time to tap on the position.

5. Nobody is choked unconscious. Not as a demonstration, not as a rite of passage, not by private agreement between two willing training partners. Loss of consciousness is never a goal at Rogue. There is nothing you can learn from being put out that you cannot learn from being in the position and tapping.

6. New people do not get pressure. Beginners learn to recognise the position and to tap, first. Pressure is added only once tapping is automatic and control is obvious. Nobody puts a strangle on at full pressure on somebody who has not yet learned to get out from underneath it.

7. My word ends it. If I stop a roll, it is stopped. You do not have to agree with me at the time and you will not get an explanation on the spot. Ask me afterwards and you will get one.

8. Tell me first. If you have high blood pressure, a heart or circulation condition, a history of stroke or TIA, a connective tissue disorder such as Marfan or Ehlers-Danlos, previous neck surgery, a stent, or you take blood thinners; tell me before you train, not after. It does not exclude you. It changes what we do.

The same goes for existing injuries. Plenty of people here train with them. It means some positions get modified or dropped, and that is treated as an ordinary part of training rather than a problem: we look at the ninety-nine things you can do and build a game out of those.

If somebody does go out

It should not happen. It has to be planned for anyway.

1. Release immediately and support the head.

2. Roll them onto their side. Clear the airway. Check that they are breathing.

3. Do not slap them, shake them, sit them up quickly, or lift them by the collar.

4. Talk to them. Expect confusion for a minute or so. Brief jerking may happen and stops on its own; it is not a seizure.

5. They are finished for the day. No more rolling, no "I'm fine".

6. Tell the instructor. Every time, without exception.

Call 000 immediately if they do not come round quickly, if breathing stops, or if there is any weakness, slurred speech, facial droop, vision change, or severe neck or head pain. Tell the operator that neck compression was involved.

And this one matters more than any of the above, because it applies to people who never went out at all. If you develop a bad headache, neck pain, slurred speech, weakness down one side, a drooping face or a change in your vision in the hours or days after training, treat it as an emergency and say that you had been grappling. A tear in an artery in the neck can show up long after everyone has gone home. It is rare. It is also the thing that gets missed.

What we don't do here, and why

Every one of these is a safety decision with a reason behind it.

We don't teach escapes from fully locked submissions. Most of the answers are hazardous. Unless there is a title and a large amount of money on the line, it is not worth the risk in a club.

White belts and blue belts do not learn leg submissions. They learn leg controls, and they learn which mechanics put a limb in danger. Heel hooks, foot locks and toe holds work on the leg the same way the Americana, the arm bar and the Kimura work on the arm; so the mechanics are learned on the upper body first, where the signal arrives earlier, the failure is less catastrophic, and the recovery is shorter.

Some positions are practice-only. The Z-lock, the leg entanglement from outside ashi garami, where the knee is internally rotated and the ankle levered up, puts enormous pressure on the knee. It is drilled in controlled practice so students understand its danger, and it is never used in free rolling.

No judo throws. There is no sprung floor, there is a roomful of existing injuries, and the return on breakfalls does not justify the cost.

No blast doubles. The room is not twenty years old.

White belts do not contest takedowns to the ground. They do not yet have the experience to do it safely, or to know what will hurt their partner.

Standing practice is partitioned. Everything is trained up to the balance break, and then it stops. The takedown itself, which is where the injuries happen, is excluded, and work resumes from top turtle.

And the injury that proves the rule. The only injuries at Rogue in recent years came from two white belts starting a free roll from standing without being clear on what they were contesting: standing only, or through to the ground. The cause was ambiguity about the rules of the exchange, not the techniques themselves. Define the contest before it starts.

Chokes and children

This is the part parents come here for, and the honest answer is the opposite of what most people expect.

The technique that worries me in a junior class is not the choke. It is the armbar.

A joint lock gives a child no warning. There is no pain signal that arrives early enough to be useful. Nothing that says stop, right up until the moment something has already gone. Children ignore shoulder locks and armbars while they are rolling, not because they are tough or careless, but because their body has not told them anything yet. By the time it does, it is too late to be a lesson.

A choke tells them straight away. The feedback is immediate, unmistakable, and it arrives a long way before any harm is done. The child feels it, the child taps, and the roll stops. That is a working safety system, and it is the reason chokes are among the first things I teach; to children and to adults alike. Not because they are advanced, but because they are the safest place to learn to tap.

The cross-lapel choke is my example, and we teach it from both mount and guard. It is relatively safe, it is hard for a child to get very tight, and it gives clear cues to tap. It is not in the class to win anything. It is in the class to build the reflex that makes everything else safe.

Front chokes are out in Kids. Not supervised, not occasionally, out. A front choke applied incorrectly puts pressure on the front of the neck, and applied by a child of that age it will be applied incorrectly. Every time. That is not me saying children are careless; it is me saying that correct application is beyond them at this age. A technique whose safety depends on precision a child does not yet have is a technique that does not belong in the room at all, and supervising it more closely does not fix that.

Youth is where that begins to change, slowly and on my terms. As control and judgement develop, a wider range of neck attacks is introduced one at a time, under my direct supervision. Anything introduced that way is drilled in controlled practice long before it is available in free rolling, and some of it stays practice-only for a long time. I decide when, and for whom. That is not a decision a student gets to make by feeling ready.

So the position is narrower than "chokes are safe for children", and I would not want you to read it that way. It is this: specific chokes, chosen because they teach the tap, with anything that loads the front of the neck excluded outright in Kids, and introduced only later and under direct supervision in Youth. Every rule in the sections above applies in both classes exactly as it applies to the adults: tap early, release instantly, nothing on the windpipe, and nobody is put out, ever.

Kids (ages 5 to 9) and Youth (ages 10 to 15) each run once a week on a Tuesday afternoon, back to back, and I teach both. The cap of twelve on the mat applies to them the same as to everybody else.

Before your first week

You will not be surprised by anything. That is the point of how the first week is set up.

Your free first week is spent in the sessions I teach; you meet the head coach at every one of them, at a spread of times, before you decide anything at all. Register in advance; Rogue does not take walk-ins. Everybody acknowledges the code of conduct before stepping on the mat. There is a hard cap of twelve people on the mat, and that includes anybody training a free week.

In those first sessions you will learn what the position looks like from the outside and from underneath, and you will learn to tap. Nobody will put a strangle on you at pressure while you are still learning that. You will spend more time being shown how to recognise trouble early than how to cause it.

We do not compete and we do not run a competition class. That removes the single biggest source of pressure to hold something a second longer than you should.

The long game

The easiest way to have a long, slow journey in jiu-jitsu is to spend it off the mat recovering from injuries.

You can take your black belt from fifteen years to ten purely by not getting injured; by being on the mat more. That is the whole argument for everything on this page.

Tapping is not losing. It is the price of being here next week.

A note on the references

Six papers are cited above. I read the Hasegawa review, the Rowe and Wedlake article and the Stellpflug 2020 paper in full. For Stacey (2021), Stellpflug (2025) and Singerman (2026) I could obtain only the published abstracts, and I have deliberately not made any claim about those three that goes beyond what their own abstracts state. I am telling you that because an earlier version of this page cited work that nobody had opened, and I would rather be dull and accurate than impressive and wrong.

Hasegawa ME, Rimm JB, Ishikawa KM, et al. (2026). Common grappling submissions: a descriptive, illustrative and literature review of anatomic structures at risk and pathophysiology. Hawai'i Journal of Health & Social Welfare 85(6), 132–141.

Rowe M & Wedlake L (2009). The carotid choke: to sleep, perchance to die? Journal of Asian Martial Arts 18(3). Republished open access in Revista de Artes Marciales Asiáticas 4(4), 50–69. doi:10.18002/rama.v4i4.151

Singerman KW, Bird CL, Cho S, Kavookjian HL & Kraft SM (2026). Laryngopharyngeal trauma in grappling martial arts: a mixed-methods study. OTO Open 10(2). doi:10.1002/oto2.70256

Stacey BS, Campbell Z & Bailey DM (2021). Elevated cerebral perfusion and preserved cognition in elite Brazilian Jiu-Jitsu athletes: evidence for neuroprotection. Scandinavian Journal of Medicine & Science in Sports 31(11), 2115–2122. doi:10.1111/sms.14031

Stellpflug SJ, Dalrymple KA, Stone D, et al. (2025). Impact of repeated sportive chokes on carotid intima media thickness and brain injury biomarkers in grappling athletes. The Physician and Sportsmedicine 53(1), 18–26. doi:10.1080/00913847.2024.2366154

Stellpflug SJ, et al. (2020). There is more to the mechanism of unconsciousness from vascular neck restraint than simply carotid compression. International Journal of Neuroscience 130(1), 103–106.

The account of police neck restraints comes from the Queensland Police Service announcement of 14 April 2023, as reported by the ABC.

Mike Armstrong. Head coach, Rogue BJJ Studios, Lyneham.

I am not a doctor. Nothing on this page is medical advice and it is not a substitute for advice from somebody qualified to give it. If you have a health condition that might be affected by any of the above, talk to your GP, and talk to me.