Crossing your hands is not a driving test fault. Here is what the RSA actually marks on steering.

Two documents govern this: the 29-page Driver Tester Marking Guidelines, published as Making Your Mark, and the 44-page Standard Procedures, Training Guide and Health & Safety requirements for conducting driving tests. Here is the whole of the first document's definition of a steering fault:

"Steering: having both hands off the steering at the same time or steering unnecessarily with one hand for a prolonged distance or over steering \ under steering." RSA Driver Tester Marking Guidelines, Vehicle Controls section. The backslash is as printed.

That is the whole of it. The guidelines present it as examples rather than an exhaustive code, so take it as the three behaviours they actually name: both hands off at once; steering unnecessarily with one hand for a prolonged distance; and over or under steering, meaning the wrong amount of input, which we come back to below.

Now the absence, which we checked rather than assumed. In the marking guidelines, "crossing hands" and "crossing arms" appear nowhere. The words "arms" and "thumb" appear zero times in 29 pages. "Cross" appears 26 times, every one of them a cross junction, a pedestrian crossing, a white line, or crossing the path of another vehicle. "Hand" appears 39 times: left-hand and right-hand positions, the handbrake, hand signals, and the two in the sentence above. The Standard Procedures are just as quiet, all five mentions of steering sitting in technical-checks lists. Neither document describes a grip or a hand position anywhere.

Compare that silence with how the same list treats the control beside it. For the handbrake the RSA spells out "being applied before the vehicle has come to a halt, or inadequate application, or attempting to drive while the handbrake is still engaged". They are perfectly willing to be specific. On steering they wrote one sentence and stopped.

We got this wrong too

Until 26 July 2026 this website told learners: "Do not cross your arms (cross-arm steering) when turning. Use the push-pull technique." Worse, the same claim sat in the fault list behind our mock test and our test result analyser, where it read "crossing hands on full lock". Learners took printed reports home with that on them. It was wrong, it is fixed in all three places, and the fix is why this article exists. Pull-push is still what we teach, but as a habit rather than a rule a tester enforces. Short version on our test faults page.

Where steering sits in the marking system

Steering gets no heading of its own in the guidelines. It is one of seven items under Vehicle Controls: "An applicant should make proper use of the: accelerator, clutch, gears, footbrake, handbrake, steering, and secondary controls or a fault may be recorded for 'Vehicle Controls' as appropriate." So the wording that applies to steering is primary-control wording. Aspect 14 of the RSA's reference chart:

The bottom band goes unwritten: "All faults the occur between 0% to 35% on the marking scale will not be recorded as they are not deemed a serious fault." (The typo is in the original.) Failure then arises from 1 or more Grade 3 faults, 4 of the same Grade 2 fault for a single aspect, 6 or more Grade 2 faults under one heading, or 9 or more overall. Full arithmetic in why did I fail my driving test.

So for steering to be marked at all, a tester has to judge it as serious misuse of a primary control. A tidy hand cross on a right turn, car where it should be and pace unchanged, does not reach the band that gets written down.

It is on display for a while, mind. The Standard Procedures set a floor: on a category B test, "in no circumstances should the time spent driving on the road be less than 25 minutes". A hand position you can only hold while concentrating will not last that long, so whatever you do with your hands has to be the thing you do without thinking.

The kerb proves the point

Here is the part that settles it, because it is the RSA drawing the line rather than us. Faults are "not 'double marked', and where 2 (or more) faults suggest themselves, it is usually the fault which conveys the most pertinent information to the applicant which is recorded." Their illustration is the exact scenario learners lie awake over:

"An example of the above would be where an applicant, e.g., hits a kerb while making a left turn. A fault may be recorded here for 'Position turning Left,' and not for 'Steering' in this case." RSA Driver Tester Marking Guidelines, Notes on the Driving Fault Marking Guidelines

Confirmed downstream: "Hitting or mounting a kerb" is the first item named in the list of incorrect Position Turning Left faults. So the loudest consequence of poor steering is not filed under steering at all. The tester records outcomes, under whichever heading tells you the most useful thing, which is why counting hand movements is the wrong model of the test.

Most steering problems we see on lessons are seating problems in disguise. The method below is Roadcraft, the UK police driver's handbook, in our own words. It is not Irish law and not an RSA requirement, just a well-regarded technique manual.

1. Shoulders back into the seat, and they stay there for the rest of this.
2. Arms straight out, wrists resting on the top of the rim. If they cannot reach with the arms straight and the shoulders still touching, you are too far back.
3. Let the arms drop. Your hands land on the sides of the wheel, between ten to two and quarter to three depending on how that wheel is moulded. That is your hold: the seat chose it, not you.
4. Cross-check twice. Clutch flat to the floor with a slight bend left in the knee, elbows slightly bent with both hands on the rim. A straight leg or a locked elbow means too far back.

The Irish bit: the seat is in the test, the technique is not

Before you move off, the tester reads a fixed script, and the seat is in it:

"Thank you. Now, would you please ensure that the seat, seat-belt, head restraint, and mirrors are properly adjusted and that your door is properly closed." RSA, Standard Procedures, Training Guide and Health & Safety requirements for conducting driving tests, cover dated 20 February 2026

Those adjustments appear on the marking chart under aspect 1, Rules/Checks, which "includes the door, mirrors, seat, seatbelt, head restraint adjustments as per the Preamble". A maximum of one Grade 2 fault may be recorded there, and the Grade 3 column reads "Not applicable". So adjusting the seat is inside the assessment, while the position you end up in is prescribed nowhere. The tester tells you to adjust, not where to finish.

Why a seat too far back manufactures a fault that is marked

Seat too far back, so the arms go nearly straight just to reach the wheel. Straight arms cannot feed a wheel, because feeding needs elbow travel and there is none left. So the only comfortable arc is across the top, and to get there the shoulders peel off the seat back.

What comes out is a recognisable pattern. One hand draped over the top of the rim, the other in the lap. Corrections that arrive late and then get overdone, a slow wander in the lane between them, whole turns made one-handed because the second hand never came back. Read that against the RSA's sentence: "steering unnecessarily with one hand for a prolonged distance", and "over steering \ under steering". Two of the three real faults, produced by a seat in the wrong place rather than by bad hands.

Seat setting belongs to Lesson 1, Car Controls and Safety Checks, of the 12 mandatory EDT lessons. It is usually the fastest four minutes of the syllabus, because everyone in the car wants to get moving, and it quietly decides what your steering looks like for the next year.

Palms on the rim, thumbs laid along it, thumbnails facing you. Not wrapped through the wheel, not hooked round a spoke, not resting on the centre boss. Grip lightly and tighten only when you need maximum control. The reason, from Roadcraft in our own words: if a front wheel hits a kerb, the wheel can spin sharply, and a thumb wrapped round it goes wherever the wheel goes. That is steering kickback, and its severity depends on the angle of the hit, the state of the suspension, the speed and the type of steering mechanism.

Why that matters in Ireland rather than in theory: look at the road edges. Cork County Council alone paid out €3.1m in public liability claims over twelve months, 86 per cent of the claims it paid from the start of 2024 to September 2025 were pothole-related, and it logged 5,338 pothole complaints between 1 June 2025 and 9 July 2026.

The evidence tier, honestly: nobody has run a trial on thumb position. What exists is a mechanism plus clinical case reports. Stoel and colleagues (2014) published two cases of airbag-induced thumb avulsion, the deploying bag forcing open the thumb webspace and tearing ligament, bone, muscle and nerve. Mahmoud and colleagues (2024) report that roughly 19 per cent of airbag-related upper-extremity injuries involve hand fractures, and name the hand caught in the wheel spokes as one mechanism.

A 2006 EMS World piece by a medic and airbag investigator adds two state troopers with thumbs fractured while resting them over the centre airbag cover, though that is trade press and case observation rather than a study. Enough to justify a free change to how you hold a wheel. Not enough to tell anyone it will save their life.

The airbag, the real argument against the top of the wheel

If there is a safety case for keeping your hands out of the top half, it is a crash-injury case, not a driving-test case. Roadcraft's warning, in our words: moving your hands past 12 o'clock puts your arms across the wheel, which increases the risk of injury to hands, arms or face if the airbag inflates. Increased risk, not a prohibition.

The published backing: Atkinson and colleagues (2002) found rising forearm fractures from driver-airbag interaction in US crash data for 1995 to 1999, and in nine cadaver experiments in the worst-case grasp, dual-stage deployments fractured two forearms while single-stage deployments fractured none of the contralateral arms tested. Mahmoud and colleagues (2024) report upper limb injury as up to four times more likely where an airbag deploys. Stoel and colleagues name hand position at deployment as the main predisposing factor. NHTSA is reported as recommending 9 and 3 on the same logic, though we did not open the NHTSA page itself.

Pull-push, briefly: neither hand passes 12 o'clock, one hand grips and makes the turn while the other slides down its own side ready to take over. Start with a pull, because a pull gives better control. Straighten by feeding the wheel back through your hands, and never let it spin back on its own.

The honest reason we teach it, which is better than the false one:

• Both hands stay on the rim, so fault one becomes very hard to commit by accident.
• There is never a reason to drape a hand over the top, so fault two goes with it.
• Neither hand crosses the top, so you can reverse the steering instantly at any point in the turn. That is what saves you when the amount of steering turns out wrong, which is fault three. Roadcraft calls that ability a feature of most safe and efficient steering techniques.

Note what Roadcraft does not do. It does not mandate pull-push. It calls it the most widely adopted of a range of police techniques, and says the method should be judged by the control, efficiency and comfort the driver gets across the full range of movement, which varies with steering weight, wheel diameter, turns lock to lock and the driver's build. Even the police manual presents it as the best default, not a rule.

And if a video told you otherwise: the DVSA in the UK publishes "crossing your hands when steering" on its own list of driving test myths, and says the examiner assesses whether your steering is smooth, safe and under control.

Where feeding the wheel is the right tool

Technique is task-dependent, and the manoeuvres are a different task. On the reverse and the turnabout you are at walking pace with your head turning, so your hands have to work while your eyes are elsewhere. Feeding the wheel through your hands is the tool for that, and Roadcraft reserves rotational steering for exactly these exceptional cases: very slow manoeuvring, very high speed, and skidding.

Two Irish specifics back that up. Under Reverse, the guidelines require direct observations on a category B car test, defined as "taking appropriate observations to the front, sides, and rear of the vehicle, including blind spots", so your head has to be moving. And on the turnabout, "a fault should not be recorded where an applicant turns the steering wheel while the vehicle is stationary (i.e. 'dry' steering)", though learners are told the opposite constantly. Neither is a licence to be sloppy: both hands off at once is still both hands off.

Over-steering and under-steering, which are not what you think

The third fault is the one nobody explains, and the word is a false friend. On an RSA report it means the wrong amount of input. Too much: swinging wide out of a left turn, or sawing at the wheel to hold a lane. Too little: running out of turn, straddling a lane, clipping a kerb, swan-necking a right turn. Amount, not method.

It is not the vehicle-dynamics meaning of understeer, where the front tyres lose grip and the car runs wider than the steering asks for. Roadcraft's advice there is counter-intuitive: when the front is running wide, adding more steering makes it worse, because you are asking for grip the front tyres do not have. More on that in the physics of braking and cornering. When an instructor says "you are under-steering" on a test route in Finglas, they mean the first thing.

1. How often steering is actually marked in Ireland. The RSA does not publish a breakdown of recorded faults by aspect, so nobody has that number, including the schools that quote percentages at you. Where this site shows a "failure frequency" bar next to a fault, that is an instructor's judgement of how often we see it, not a measured national rate.

2. Whether pull-push is safer than rotational steering. No trial has compared them for crash outcomes. The document most often put in front of instructors as proof is an advocacy essay by a retired Metropolitan Police class 1 instructor, hosted on a RoSPA county website, written as a rebuttal of Sir John Whitmore and dating from January 2010 by the PDF's own file date. We downloaded it and read all ten pages: anecdote, four web addresses of which two are now dead, an unsourced airbag deployment figure, and a quoted email from a retired LAPD officer. No study, no data table, no comparison of methods. An argument, possibly a correct one, but not evidence.

3. Formal driver training has a poor record on crash outcomes generally. Pooled across fifteen trials, post-licence driver education came out at a relative risk of 0.98 for traffic crashes, 95 per cent confidence interval 0.96 to 1.01. No detectable reduction. Anyone telling you a hand technique saves lives is ahead of the evidence, and we have written about that at length because it is uncomfortable for our own industry.

4. Testers are people. The guidelines say so: they "do not purport to cover every possible situation which can arise, and they may be added to, or amended from time to time", and testers work from them plus guidance from their Operations Supervisor and their own experience. So do not walk into a test with a lawyer's argument ready. Steer well enough that the question never arises.

1. Set the seat by the wrist test before you touch the key. Every drive, including in a car you have driven a hundred times.
2. Clutch flat to the floor, knee still slightly bent. If the leg goes straight, come forward a notch.
3. Thumbs along the rim, never hooked through the wheel or a spoke.
4. Both hands on the rim. One comes off to work a control and goes straight back. That habit alone deals with two of the three real faults.
5. Pull-push on the road, feeding the wheel on the reverse and the turnabout. And stop apologising for the second one.
6. Look where you want the car to end up, not at the kerb you are afraid of. Your hands follow your eyes, so this fixes more steering than everything above it combined.

If steering unravels under pressure, another lap of a quiet estate will not tell you much. A full mock test will, because that is where you find out what your hands actually do when you are nervous and watched. Cheaper than finding out on the day, in a test centre car park, on a booking you waited months for.