Knife Grips Explained: Forward vs. Reverse for Real-World Self-Defense

What the grip debate gets wrong, what each grip is about to do to you, and why this matters more for defending against a knife than for holding one.

Jared Wihongi, Grand Tuhon of the Pekiti Tirsia Tactical Association, gets asked how to hold a knife often enough that he has a stock answer ready. "Tightly."

It reads as a joke, and it is one, but there's a real argument buried in it. Every version of the grip debate you've seen online assumes you get to choose. Clean stance, sensible distance, a moment to think. That situation exists in demos and almost nowhere else. In civilian violence you don't select a grip. You land in one, usually badly, usually while someone has a fistful of your hoodie.

So there are really two versions of this question. If you carry, it's about what your hand will find when you need it, and you settle that long before anything happens. If you don't carry, which is most people, it was never your question in the first place. It's about his grip and what it's about to do to you.

Much of what follows is the second one.

What a knife attack looks like

Civilian knife attacks are rarely a square-up. What shows up over and over is an ambush at contact range. One hand anchors you, usually the collar or the sleeve or the backpack strap, sometimes the hair or the neck. The other hand works underneath in short violent bursts.

Two shapes recur, and the grip tells you which one is coming. The first is the sewing machine: forward grip, short straight thrusts to the abdomen, low ribs, and hip crease, in and out on the same line, as fast as he can work, his weight behind every one. The second is reverse grip, descending arcs onto the head and chest, thrown again and again from above.

The detail that matters most is that you may very well feel it before you see it. The grab lands first. The blade is often hidden until it isn't. Anything built on the assumption that you spot the weapon at ten feet and make a considered decision is solving a problem you probably won't have the luxury to solve.

The grips, and what each one is telling you

It’s worth learning the words. Not for the sake of the words, but because naming the grip is the fastest way to know roughly where the blade is about to travel. I'm describing each one from the point of view of the person holding it, because what it does for them is exactly what tells you what it's about to do to you.

Forward grips. The hammer grip (saksak to Kali practitioners) is a neutral fist with the blade forward, the way you'd hold a hammer. It's the most common thing you'll see, it retains well when things collapse to the body, and it's easy to hang onto in a scramble. The saber grip puts the thumb on the spine or ramp, which buys a little reach and finer tip control, but only if there's a clean lane to use it in.

Reverse grips. Icepick (pakal to Kali practitioners) points the blade down out of the bottom of the fist. There are two versions, depending on which way the edge is turned. Edge-in, with the edge toward the attacker's own body, favors hooking and withdrawing lines in entanglement and retains very well. Edge-out favors downward and backhand arcs that scrape along the outside of a limb.

The tradeoff is simple. Reverse grips lose reach against forward grips, and gain leverage and retention once you're chest to chest.

For a defender, that's the whole payoff. Forward grip means longer reach than you'd guess and pressure arriving mostly on diagonal or horizontal plane in front of you. Reverse grip means shorter reach, arcs coming down and around, and an attacker who is comfortable being close to you. Same threat, two different shapes. What stops one won't necessarily stop the other.

Which hand, which lead

This is the part of Kali that's useful even if you never plan to hold a knife. Pekiti Tirsia spends a lot of time on angles, on where a blade goes and why. Do enough of that from the attacking side and the defending side stops looking like chaos.

Some of what you learn on the attacking side transfers straight over. A knife in the front hand reaches you fastest and is the easiest to see coming. A knife in the back hand is slower, hides better, and leaves his front hand free to grab you first. They are both worth worrying about.

Most of the time the knife is in the strong hand and the grab is in the other one. Right-handed guy, blade on the right, grabbing you with his left. Train both sides anyway. You'll guess wrong sometimes, and you don't get to pick which side he comes from.

If you want to go further with this, two people are worth your time. Craig Douglas and the ShivWorks crew have done some great public work on what happens to technique once you're tangled up with someone, which is most of the time. And Michael Janich, through Martial Blade Concepts, is the guy who made a lot of us think harder about where the damage happens on a body. He mostly argues the other way from the reverse grip crowd, which is useful, because reading both sides is how you stop treating your own preference as the answer.

Getting to the weapon arm

Here's the part that sounds like a contradiction. The answer is the same whatever grip they are holding. Get two hands on the weapon arm. That's what ShivWorks teaches, it's what Ryan Hoover teaches, it's what Knife Control Concepts teaches, and it's what we teach. None of that came from the big legacy Krav Maga systems. It came from grappling, and from people who took it seriously enough to pressure test it against a blade. 

Naming the grip doesn't change what you do. It changes what you expect: how much reach he has, where the next one comes from, how much of his weight is behind it. You're not choosing a different technique. You're arriving at the same one less surprised.

First you need to work out which hand the knife is in. Once you’re attached to the weapon bearing limg, the job is to stop them driving forward. That means a wide stance, your head driving up under their chin, arms locked, and the weight of your chest and head sitting heavy over the weapon. In our clinch training we call that a baseball bat grip, or a frontside two-on-one. 

You won't get it cleanly. It's a fight to get there and a fight to keep it, and you'll likely take some damage on the way in. That's the honest version and it's still the best thing available. The alternative is staying unattached and trying to strike or move your way out while the blade is free to keep working, which is worse in every direction. Being on the weapon arm and struggling beats being off it and hoping.

It isn't a fix. It's a starting point. From there you have to read fast what Paulo Rubio calls the totality of the circumstance. Are you by yourself, with somewhere to run? That points you one direction. Are you covering a kid or a partner, in a space you can't get out of? That's a different set of choices entirely.

In the room, that means training from contact instead of from distance. 

Start already grabbed. Keep it simple, just a shirt grab. From there the attacker runs a low sewing machine, a repeated set of thrusts to the abdomen with a forward grip. The defender's job is to secure the two-on-one. First priority is keeping control of the weapon arm. Then improve your position, usually to the outside or the back. Then either take them down or get out.

Once that's working, make it harder and perhaps more realistic. Put it against a wall or in a corner. Put furniture in the way. Add striking. Add a third person.

Fair warning: this is not flattering training. The defender loses a lot, and people who are used to clean reps find it demoralizing at first. That's the point. If the drill always works, you've built it wrong.

Safety first, obviously. We strongly recommend soft round-tipped training knives (we like Tak, Bunal Brand, Aku Strike but there are a ton of great options) and eye protection.

Then debrief. Two or three minutes with your partner or your coach on what worked and what didn't. The dynamic part is what makes it real, but the reflection is where the skill gets built.

Five things people get wrong

First, "reverse grip is always best." Not true. It depends on where you are. Up close it's harder to strip out of your hand, which is worth a lot. Further out you give up reach and you lose sight of what your own blade is doing. There's no setting where one of them wins everywhere.

Second, "edge-in is the only real way." There's a whole internet argument attached to this one. Skip it. Edge-in rips on the way back out, which is useful when you're already tied up with someone. Edge-out cuts on the way through, better for arcs coming down and around. People land in whichever one they've trained and then argue backwards from there. For you it doesn't matter which is better. It matters that you can tell them apart, because they arrive on different lines.

Third, "real fights are duels." They aren't. Almost every civilian knife incident starts messy and close, usually with a hand on you before you've seen anything. If your training starts at ten feet with both people squared up, you're rehearsing something that mostly doesn't happen.

Fourth, "any folder works the same." Not really. Where the clip sits, which way the knife rides, how the handle fills your hand, all of it decides what your grip looks like by the time the knife is out. Whatever you carry, carry the same way every time and put reps into that draw specifically.

Fifth, "once I pick a grip, I'm set." You didn't pick it. That's the whole point of this post. You got grabbed, your hand found the handle, and whatever it found is what you're working with. Holding onto it and moving your feet will do more for you than any amount of deciding.

If you carry

I'll keep this short, because carrying a knife is a whole other conversation and this post is about defending against one.

Getting to the knife and keeping hold of it beat having the perfect grip. Centerline and front-pocket carry have a real advantage under ambush, which is that if one arm is pinned by the anchor hand, the other may still reach the knife. So train the other hand too. Not because you'll ever want to draw left-handed, but because the day your right arm is pinned to your chest is exactly the day you'll need to. And practice opening one-handed in bad positions: shoulders jammed, hips turned, back against a vehicle door.

Then there's the knife itself: clip position, which way it sits in your pocket, whether it opens as you draw, how the handle fits your hand. Those choices do more to determine your grip than anything you decide in the moment.

Legality, safety, and what happens after

None of this is legal advice, and California knife law is more specific than most people who carry assume. Concealed fixed blades are out. Switchblade rules turn on blade length. Public buildings, schools, and airports each add restrictions, and San Francisco has local code on top of that. Using force runs on imminence, reasonable belief, and proportionality. Our legal primer goes through all of it properly.

If you don't understand both the law and the knife you're carrying, don't carry.

One more thing the drills rarely cover. Once you're out, call 911, get first aid moving, and call a lawyer before you explain yourself to anyone. Then stay out. Going back for your phone, for your bag, to check on the guy, is how people who survived turn it into something else.

Where Krav gets this wrong

All of this rests on one idea. The grip decides what the defense has to solve. Someone driving a forward-grip thrust into your hip crease and someone bringing a reverse-grip arc down onto the outside of your arm are not doing the same thing to you. Different line, different range, different amount of their weight behind it.

Krav Maga handles that part well. The overhead reverse grip, the straight thrust, the low circular sewing machine, they're all in the curriculum and they're taught as distinct problems. That's not where the gap is.

The gap is what the sequence is built for. Krav has an answer here: redirect, control, attack, take the weapon away. The pieces are fine. But the whole thing is organized around staying unattached. Get off the line, redirect, strike, strip, leave. Control is a beat you pass through on the way out, not a place you fight to reach and fight to keep.

That's the opposite of what I just spent four paragraphs arguing. And it shows up in how we drill. We run the attack, the defense, the escape, then reset. Clean start, clean finish, nothing before it and nothing after. Almost no roleplay, and very little work on pre-assault cues, which is where most of your decisions actually get made in the incidents that aren't pure ambushes. We tend to pick the ending in advance too, so you either bail or you stay and fight, but you rarely have to work out which one this is while it's happening. Each piece gets trained well and separately, and the seams never get tested. Let alone “going to the ground” - the antithesis of much Krav Maga teaching.

Choreography isn't the enemy here, and neither is drilling a situation. Slow, technical, repetitive reps where you succeed over and over are most of how anyone gets good at anything. The problem is a narrow set of situations paired with an oversimplified set of answers, held in place by a lot of legacy Krav dogma about how it's supposed to be done.

That's a bigger argument than I can finish at the end of a post about grips, and I'll make it properly soon.

The actual answer

Both grips work, and both cost you something. Reverse trades reach for retention. Forward trades the other way. People will argue about that trade for as long as there's an internet, and neither side is exactly wrong.

What matters is that you probably aren't going to make the trade. If you carry, you made it weeks ago, when you picked a knife and a pocket and started drawing it the same way every time. If you don't carry, which is most of you, it was never yours to make. It belongs to the guy in front of you, and there's a decent chance he didn't make it either.

So the skill isn't picking. It's naming. See the grip, know the line, get both hands on the arm.

Want to pressure-test this with people who'll go hard enough to be useful and safe enough that you come back? We’d love to train with you.