The first time somebody explains countersteering to you, it sounds like a trick question. Push the right grip to go right — except pushing the right grip points the wheel left. Half the internet says you steer the opposite way. The other half says you press toward the turn. Both halves are describing the same hand doing the same thing, and the argument has been running for decades.
So I went to the documents instead. Two official rider manuals — the booklet the Motorcycle Safety Foundation writes and American states hand out, and California’s own handbook — plus a peer-reviewed physics paper that actually put numbers on the forces, and a Science paper that built a bicycle specifically to kill the explanation everybody repeats.
Between the two manuals there are 118 pages and 38,063 words of official instruction on how to ride a motorcycle in America. The word appears once. And it is on the page about a vehicle that cannot do it.
The one page that names it is the page about three wheels
The Motorcycle Safety Foundation’s Motorcycle Operator Manual, 18th edition, is 58 pages and 20,349 words. It is the book the licence programmes are built on, and the source I went through page by page for the guide on learning to ride.
Search it for “counter” and you get four hits. One is counterbalance, in the section on slow tight turns. One is encounter. One is over-the-counter, in the section on alcohol and drugs. The fourth is the only occurrence of the word itself, and it is in the Three-Wheel Supplement at the back:
Because conventional three-wheel motorcycles cannot lean, they cannot countersteer. Instead, the front wheel is pointed in the direction the rider wants the motorcycle to go.
Read that again with the rest of the manual in mind. The supplement starts on page 46; in the 45 pages before it, all of them about two-wheeled motorcycles, the technique is never named. In the supplement about trikes and sidecars, it gets named exactly once — to tell you it does not apply.
That is not sloppiness. It is a deliberate choice about what a learner needs, and the second half of the sentence explains the choice: on a trike, the front wheel is pointed where you want to go, which is what everyone assumes a two-wheeler does too. The manual only reaches for the word at the moment the assumption becomes true.
What the manual teaches instead is a verb: press
The two-wheel instruction lives in a panel headed NORMAL TURNS, and it is four steps — Slow, Look, Press, Roll. The third one is the whole subject of this article, written without its name:
PRESS — To turn, the motorcycle must lean. To lean the motorcycle, press on the handgrip in the direction of the turn. Press left handgrip — lean left — go left. Press right handgrip — lean right — go right. The higher the speed in a turn, or the sharper the turn, the greater the lean angle needs to be.
Notice the order of the argument. It does not start with the handlebars. It starts with the lean: to turn, the motorcycle must lean. The handlebar is introduced as the tool that produces the lean, not as the thing that aims the bike. Everything that confuses people about this subject comes from getting those two backwards.
On the exam, “steer the opposite way” is the wrong answer
The knowledge test at the back of the manual has five sample questions. The third one is this, verbatim:
3. To swerve correctly:
A. Shift your weight quickly.
B. Turn the handlebars quickly.
C. Press the handgrip in the direction of the turn.
D. Press the handgrip in the opposite direction of the turn.
The answer key on the following page gives the correct answers as 1-B, 2-C, 3-C, 4-C, 5-B.
So the right answer is C, and the trap laid for you is D — which is, word for word, how most of the internet defines the very technique the question is testing. “Steer opposite” is offered as the distractor.
The two descriptions are not in physical contradiction. Pressing the right grip forward is rotating the bar to the left; the hand and the wheel are doing exactly what both camps say. The contradiction is a teaching one, and the manual has picked a side. It describes the input the rider gives, not the rotation the wheel makes, because the rider can only control the first one. Someone who has memorised “steer opposite” and then has to swerve at short notice has one extra inversion to perform under pressure, and the exam is built to catch exactly that person.
California’s handbook is 60 pages long and never uses the word at all
California does not hand out the federal booklet; it writes its own. The California Motorcycle Handbook, DL 665, revised April 2016: sixty pages, 17,714 words.
Occurrences of “countersteer”: zero. Apart from two uses of “encounter”, the only “counter” in the whole handbook is counter-clockwise, describing which way to ride round a circle in a skill exercise.
And yet the instruction is nearly word for word the same:
3. Press—To turn, the motorcycle must lean. To lean the motorcycle, press on the handlegrip in the direction of the turn. Press left—lean left—go left. Press right—lean right—go right. Higher speeds and/or tighter turns require the motorcycle to lean more.
Two independent documents, one national and one state-level, 118 pages between them, both teaching the same physical act, and between them the word turns up once — about a trike.
The word is in an American licence manual. It is the one for trucks, and it means the opposite
Here is the part that makes the whole thing funny. There is an official American driver-licensing document where “countersteer” appears as a named, examinable concept, in bold, with its own bullet point. It is the Commercial Driver’s Manual — the CDL book — and this is §2.19.2, Correcting a Drive-wheel Braking Skid, from Georgia’s published copy of it:
Countersteer. As a vehicle turns back on course, it has a tendency to keep on turning. Unless you turn the steering wheel quickly the other way, you may find yourself skidding in the opposite direction.
And immediately after:
Learning to stay off the brake, turn the steering wheel quickly, push in the clutch, and countersteer in a skid takes a lot of practice.
Same word. Opposite moment. On a truck you countersteer at the end of a slide, to stop the recovery from overshooting into a second slide. On a motorcycle you countersteer at the beginning of a turn, to start it. A trucker and a motorcyclist can both say “I countersteered” about the same afternoon and be describing two unrelated actions at two opposite ends of an event.
If you have ever wondered why the term generates such reliable arguments, that is a large part of the answer: it is one word doing two jobs, and only one of the two has ever been written down officially.
What it actually takes to turn a motorcycle: 12.7 newtons
Now the number that is missing from every page currently ranking for this subject.
In July 2000 the American Journal of Physics published “Steering in bicycles and motorcycles” by Joel Fajans, of the physics department at Berkeley. It is six pages, it models the whole manoeuvre, and it works one example on an actual motorcycle: the parameters in his notes are those of a Harley-Davidson XLH Sportster 883, total mass of machine plus rider 300 kg, travelling at 20 m/s — about 45 mph.
The turn he puts it through has a 200-metre radius, and that number is not arbitrary either. His note says the radii used “are the minimum highway turning radii allowed at the given speeds” — the tightest curve an American highway is permitted to have at that speed. So: a real motorcycle, at a real speed, in the sharpest legal curve.
The handlebar effort that turn requires:
The handlebar torques used in this example are higher than for the bicycle: 12.7 N m, or a force of 12.7 N for handholds 0.5 m apart, the equivalent of a weight of 1.3 kg.
One point three kilograms. That is what it takes to send 300 kg of motorcycle and rider into the sharpest curve a highway is allowed to have, at 45 mph. Rather less than a large bottle of water, resting on the grip.
I am quoting his arithmetic rather than redoing it, because the step from 12.7 N·m to 12.7 N is his and depends on how he treats the handholds. The point survives any version of the sum: the effort is tiny, and it is tiny in a way that explains an experience every rider has had. On a bike that feels heavy at walking pace — and motorcycles weigh what they weigh — the steering effort at road speed is almost nothing. The mass you fight in the garage is not the mass you fight in a corner.
It also explains the classic beginner’s problem in reverse. If a kilo and a third of push is a sharp turn, then an unintended half-kilo of push — a tense arm, a hand braced against a bump, a shoulder tugging as you look over it — is an unintended turn. Riders who “run wide” are very often riders who are holding on.
The gyroscope is not what leans the bike, and it matters less the faster you go
Ask why a motorcycle behaves this way and you will be told about gyroscopic precession: the spinning front wheel resists being tipped, so pushing the bar makes the bike lean. It is the explanation in the diagrams, in the training-school blog posts, and in a good deal of published physics.
Fajans checked it, and his abstract is blunt:
Contrary to common belief, gyroscopic forces play only a limited role in balancing and steering.
On the motorcycle example specifically:
Contrary to the assertion in Ref. 3, gyroscopic action plays no role in leaning the bike. However, as shown in Fig. 3(d), and in agreement with Ref. 3, it does play a role in steering the front wheel back towards the desired direction.
That is a precise split, and it is worth keeping precise. The gyroscope does not lean the machine. It does help the front wheel come back around to point into the turn once the lean has started. Two different jobs, and the folk explanation has handed it the wrong one.
He then puts numbers on how small the first job is. For the motorcycle case — wheel radius 0.29 m, effective wheel mass 9.0 kg, over a one-second input — he divides the two gyroscopic terms by the centrifugal term and reports the result:
For Fig. 3, with r=0.29 m, m=9.0 kg, and τ=1 s, the ratios are 0.029 and 0.0011, respectively.
Just under 3 percent, and roughly one part in a thousand. Against the force that actually leans the bike, the gyroscope is a rounding error.
And then the line that inverts the intuition completely:
Increasing the speed only makes ωτ bigger, so, for the lean, gyroscopic action actually becomes less important at high speed.
Everyone who has been told that the spinning wheels are what hold a motorcycle up at speed has it backwards, at least for the leaning part. Faster does not mean more gyroscope. For the lean, faster means less.
The paper’s conclusion says so without hedging:
The appealing notion that gyroscopic forces are central to bike behavior, often repeated in papers and textbooks, is incorrect.
Somebody built the bicycle with the gyroscope cancelled out
Eleven years later, Science published the experiment. Kooijman, Meijaard, Papadopoulos, Ruina and Schwab set out to test whether the two standard explanations for a riderless bicycle balancing itself — gyroscopic precession and caster-like trail — are actually necessary. Their abstract states the belief they are testing:
A riderless bicycle can automatically steer itself so as to recover from falls. The common view is that this self-steering is caused by gyroscopic precession of the front wheel, or by the wheel contact trailing like a caster behind the steer axis.
Then they built a machine with both effects removed: extra counter-rotating wheels to cancel the spin angular momentum outright, and the front-wheel contact patch moved ahead of the steer axis, making the trail negative. Push it sideways while it is rolling and it still recovers to upright on its own. Neither effect, it turns out, is necessary.
The same paper states the mechanism in one clause that is worth memorising, because it is the whole of balance on two wheels:
bicycle and rider in forward motion balance by steering toward a fall, which brings the wheels back under the rider
That is a bicycle experiment, not a motorcycle one, and the difference in mass and geometry is real. But the belief it demolishes is the same belief, and it is the one you were probably taught.
The wrong-way movement everybody worries about is under six centimetres
The reason countersteering frightens people when it is first explained is the mental picture: you want to go right, and the first thing the machine does is go left. Into the oncoming lane, or into the kerb, depending on which way you are dodging.
Fajans computed that too, across every case he modelled — bicycle and motorcycle, bar input and hip input:
the bike initially always turns in the direction opposite to the desired turn. However, this initial turn results in a very small incorrectly directed deviation: less than 6 cm in all the examples shown here.
Under six centimetres. Not a lane, not a wheel’s width — about the width of your palm. It is a real displacement and it is genuinely in the wrong direction, and it is small enough that it never arrives as a movement you can feel.
That caveat matters: “in all the examples shown here” means the cases in the paper, not a universal law. But those cases include a Sportster at 45 mph in the tightest curve a highway is allowed, which is not a gentle test.
The parameter that decides how a motorcycle feels is trail, not spin
If the gyroscope is not the answer, what is? Fajans keeps pointing at one piece of geometry — trail, the distance by which the tyre’s contact patch sits behind where the steering axis meets the road:
The trail makes the bike self-steer: when the bike leans to the left, the front wheel steers left; when the bike leans to the right, the front wheel steers right. This effect is easily demonstrated by standing beside a bicycle and leaning it from side to side.
That demonstration takes ten seconds and no risk at all: stand next to a bicycle, hold it by the saddle, tip it left and watch the bar swing left by itself. Nothing is spinning. The wheel steers anyway.
His verdict on where the character of a machine comes from:
the “feel” of the bike is dominated by the trail.
and, in parentheses in the middle of the paper:
(The trail is the single most important geometric parameter which enters into the handling of a bicycle.)
Which is why a change of tyre profile or a badly set pre-load alters how a bike turns in a way that feels like a different machine, and why worn steering-head bearings make it feel vague long before anything is visibly wrong. The geometry that steers the bike is being interfered with. Tyre pressure and profile belong to the same short list for the same reason.
What one turn actually looks like, in order
Fajans breaks a countersteered right turn into five steps. Stripped of the equations, and with his own caveat that the division is “somewhat arbitrarily divided”:
- You apply a torque to the handlebars, steering the front wheel left.
- The wheel goes left, the bike starts turning left, and centrifugal torque leans you and the frame right. (Gyroscopic action leans it right as well — negligibly, as the numbers above show.)
- The growing lean tries to tip the front wheel over too. Here, for the first time, the gyroscope matters: the wheel responds by trying to steer right, working against your input, and the steering angle stops growing.
- The steering angle starts coming back — while the lean keeps increasing, because the bike is still turning left.
- With the lean established, the steering angle passes smoothly through zero and points right. The centrifugal torques reverse, the lean stops growing, gravity and cornering forces balance, and the bike settles into the right-hand turn you asked for.
The whole sequence is over in well under a second, which is why nobody experiences it as five things. But the shape of it is the reason the instruction in both manuals is press, not steer: you supply step 1, and the machine does 2 through 5 without consulting you.
There is one demonstration in the paper that anyone can run, and it is on a bicycle rather than a motorcycle — deliberately:
While riding at a brisk pace (possibly downhill to avoid the complications of peddling), let go with your left hand while pushing the right handlebar with the open palm of your right hand. Since your hand is open, you can only turn the handlebar left, but the bike will turn right.
An open palm can push but cannot pull, so it removes the ambiguity about what your hands are doing. Do it on a bicycle, in a place with room, and the argument ends.
The other half of the same physics is braking, and here the manuals get strict
Leaning and stopping draw from the same account. There is one contact patch at each end and a fixed amount of grip in it, and every newton spent turning is a newton not available for stopping. Both manuals say so plainly, and then both of them turn it into a rule.
The Motorcycle Operator Manual, in the emergency-braking chapter:
If the situation allows, separate braking from swerving.
The same booklet, phrasing the rule cleanly in the three-wheel supplement:
If braking is required, brake either before or after the swerve, never while swerving.
And California, which adds the reason:
If braking is required, separate it from swerving. Brake before or after, never while swerving, especially the front brake as this may cause the motorcycle to fall over.
Now go and search for trail braking — carrying the brake into the corner and easing it off as the lean builds. The first page of results is racing schools, sim-racing tutorials and car-track coaching. It is a real technique, it is taught seriously, and it is the deliberate overlap of exactly the two inputs both licensing manuals tell you to keep apart.
Both are right, because they are answering different questions. Trail braking is a technique for a known corner on a closed circuit, where the plan is made in advance and the grip budget is being managed on purpose. The manuals are legislating for an unplanned swerve on a public road, where the rider has no plan and no time to make one. Same physics, opposite risk profile.
The manual even contains the mirror image of trail braking, for the case where you must stop in a bend:
If you must stop quickly while in a curve, first straighten and square the handlebars, then stop.
and, when straightening first is not possible:
As you slow, you can reduce your lean angle and apply more brake pressure until the motorcycle is straight and maximum brake pressure can be applied.
That is the same traction budget run backwards: on a racetrack you trade brake away as lean comes on, on the road you trade lean away as brake comes on. Nobody escapes the arithmetic; they just walk around it from opposite sides. If your machine has cornering ABS, the electronics are doing a version of the same trade for you, and knowing what it is trading is the difference between using it and counting on it.
What the people who crash get wrong
The manual does not leave this to intuition. Just before the quick-stop instructions there is a short, unhappy list:
Studies show that most crash-involved riders:
Underbrake the front tire and overbrake the rear.
Did not separate braking from swerving or did not choose swerving when it was appropriate.
Two failures, and both are failures of the same budget. Too little front and too much rear wastes the brake that does most of the work — the manual is explicit about the split, in the braking section:
The front brake is more powerful and can provide 70% or more of your total stopping power.
That figure is on the exam as well: question 2 asks how much of the potential stopping power the front brake supplies, and the answer key says at least 70%. Which is worth remembering the next time somebody tells you the front brake is the dangerous one. It is the one that stops the motorcycle. Whether it feels like it does depends on how recently the fluid was changed — a spongy lever is a rider guessing at 70 percent of the stopping power.
The second failure is subtler and more common: not choosing. Faced with a hazard, the untrained reaction is to brake, because braking is what a car driver’s body knows. On a motorcycle, swerving is frequently the better answer, and the decision has to be made before the hands move. That is the same reflex the defensive-riding habits are built to buy time for.
Where the curve stops being theory
California’s handbook opens its cornering section with a sentence that is really a crash statistic:
A primary cause of single-vehicle collisions is motorcyclists running wide in a curve or turn causing the motorcycle to leave the roadway or collide with an object.
The National Highway Traffic Safety Administration gives that category its size. In 2023, 6,335 motorcyclists were killed on American roads — 15 percent of all traffic deaths — and 2,171 riders died in single-vehicle crashes. Running wide in a bend is not the whole of that number, and I am not going to pretend it is. It is one of the ways riders end up in it, on a road with no other vehicle involved.
The same publication adds a second number that belongs in the same paragraph: 36 percent of motorcycle riders in fatal crashes in 2023 were speeding, against 22 percent of passenger-car drivers.
And the handbook’s own remedy is a single line that has nothing to do with force or physics:
Your best path may not always follow the curve of the road.
The rider who runs wide is very often the rider who was steering along the painted line instead of choosing a line, and who then, mid-corner, discovers that the curve is tightening. At that point the only inputs left are the ones this article is about: more press, and no more brake than the lean can spare. Both manuals say the same thing about where to look while you do it — through the turn, to where you want to go — and they say it before they say anything about the handlebars. In the rain the grip budget is smaller, and every margin in this paragraph goes down with it.
What is actually worth practising
None of this is knowledge that helps you in the moment. Nobody has ever recalled a physics paper mid-swerve. What the documents are good for is telling you what to rehearse, so that the right thing is automatic when there is no time to choose it.
The manual’s own on-motorcycle skill test names the list: accelerate, brake and turn safely; stop, turn and swerve quickly; complete normal and quick stops; complete normal and quick turns or swerves. Which is a practice plan, in an empty, legal, level space, with gear on:
- Press, deliberately. At a steady speed in a straight line, press one grip and then the other, gently, and feel the machine lean and come back. The aim is not the manoeuvre; it is connecting the hand to the lean so the connection is not a surprise later.
- Swerve, then recover. California spells out the second half, which most riders forget: “To swerve to the left, press the left handlegrip, then press the right to recover.” A swerve is two inputs, not one.
- Quick stops, straight, repeatedly. Both brakes, front squeezed progressively and never grabbed, body centred, eyes up and well ahead. This is where the 70 percent gets built into your hands.
- Never both at once. Brake, then swerve, or swerve, then brake. Rehearse them as separate things so that under pressure they stay separate things.
- Slow, tight turns with the body upright. The manual’s counterbalance instruction: lean the machine, keep yourself straight. It is the opposite of what you do at speed, and it is the skill that separates riders who can turn a bike round in a road from riders who put a foot down. The clutch friction zone is doing most of the work here.
And the honest sentence, which the Motorcycle Safety Foundation writes about its own book: reading is not what teaches this. Control of direction, speed and balance is something the manual says you can learn only through practice, preferably in a formal course. It puts that in a grey box at the top of a chapter, before any of the technique — and five thousand words later I am not going to pretend otherwise.
Where RideLog fits, and where it does not
I build RideLog, and I am going to be straight about this one, because the honest answer is mostly negative.
The app cannot see a single thing this article is about. It does not measure the pressure on the grip. It does not know your lean angle. It has no opinion on your braking, and it cannot tell whether you were looking through the corner or at the pothole. A phone in a jacket pocket is not an instrument for any of that, and a log tells you how much you have ridden, not how well.
What it does hold is the one variable that actually correlates with all of this: time on the machine. Trips are detected automatically from motion and GPS, with no start button to forget, so the mileage is real rather than remembered. Every route you have ridden stays there to look at again. The statistics roll up by day, week, month and year. Several bikes are tracked separately, which matters more than it sounds — skill is machine-specific, and the bike you have ridden 4,000 miles on is not the bike you picked up in April. Everything stays on the phone: no cloud, no server, no third party, and it works offline where the roads worth practising on tend to be. Trips and fuel export to PDF or CSV whenever you want them.
That is the whole claim. The thing that fixes a swerve is an empty car park and a course with an instructor in it. The app just makes it hard to lie to yourself about how much you have actually been riding this year.
Download RideLog free — iPhone and Android, one link for both.
Frequently asked questions
What is countersteering on a motorcycle?
It is the input that makes a motorcycle lean, and therefore turn, above walking pace: a brief press forward on the grip on the side you want to go. That press rotates the handlebar momentarily the other way, which is where the name comes from. The Motorcycle Safety Foundation’s manual teaches it as “Press left handgrip — lean left — go left”, without ever using the word on a two-wheeler.
Do you press the grip toward the turn or away from it?
Toward. On the knowledge test in the Motorcycle Operator Manual, “Press the handgrip in the direction of the turn” is the correct answer and “Press the handgrip in the opposite direction of the turn” is the marked distractor. Both descriptions refer to the same hand movement, but the manual deliberately describes the rider’s input rather than the wheel’s rotation, because that is the part the rider controls.
Is countersteering caused by the gyroscopic effect?
No — not the part that makes the bike lean. Fajans’s 2000 paper in the American Journal of Physics calculates the gyroscopic terms at roughly 3 percent and 0.1 percent of the force that actually produces the lean, and concludes that “gyroscopic action plays no role in leaning the bike”, though it does help swing the front wheel back towards the intended direction. His paper also notes that for the lean, gyroscopic action matters less at higher speed, not more.
How hard do you have to push?
Very little. Modelling a Sportster 883 with rider at a combined 300 kg and 20 m/s — about 45 mph — through the tightest curve an American highway is allowed at that speed, Fajans reports a handlebar torque of 12.7 N·m, which he describes as “the equivalent of a weight of 1.3 kg”. Steering effort on a motorcycle at road speed has almost nothing to do with how heavy the machine feels when it is stationary.
Does the bike really go the wrong way first?
Yes, and by an amount nobody notices. Across every case modelled in the paper — bicycle and motorcycle, handlebar input and hip input — the initial deviation in the wrong direction is “less than 6 cm”. It is real, it is measurable, and it is roughly the width of a hand.
Is trail braking the same thing, and should I do it on the road?
They are different things that share a traction budget. Trail braking means carrying the brake into a corner and easing it off as lean increases; it is a circuit technique for a corner you already know. Both the Motorcycle Safety Foundation manual and the California handbook tell road riders to do the opposite in an emergency — “brake either before or after the swerve, never while swerving” — because an unplanned swerve leaves no margin to manage. California even gives the failure mode: applying the front brake mid-swerve “may cause the motorcycle to fall over”.
Why does the truck manual use the word and the motorcycle manual does not?
Because it means something else there. In the Commercial Driver’s Manual, §2.19.2, countersteering is a skid recovery action: as the vehicle comes back on course you turn the wheel quickly the other way to stop it overshooting into a second skid. On a truck it happens at the end of an event; on a motorcycle it happens at the start of one. One word, two jobs — which is a decent explanation of why the term causes so many arguments.
The documents behind this article are all free, and worth an hour: the Motorcycle Operator Manual, whichever handbook your state publishes, and Fajans’s six pages of physics. And if you would like the mileage they apply to counted for you while you go and practise, RideLog is free to download for iPhone and Android.
