Who goes first? The unwritten rules of the narrow-road squeeze

Researchers call it a narrow passage interaction: a stretch of road where parked vehicles, works or the road itself leave room for only one car at a time, and two vehicles approach from opposite ends. One of you has to wait, or one of you has to find a gap to tuck into, or you both have to breathe in and creep past each other. There is no signal to arbitrate it. The researchers describe it as a high-risk and relatively unregulated situation, and anyone who learned to drive in an Irish town knows exactly what they mean.

It is also one of the situations learners fear most, for a specific reason: it is a negotiation, not a procedure. A junction can be drilled. A narrow passage has another human in it, and the outcome depends on what both of you believe the other one is about to do. Get that belief wrong and the best case is an embarrassing stand-off where you both wait, both go, and both stop again. The worst case is two cars nose to nose with traffic stacking up behind each of them, or a mirror clipped at 40 km/h.

The reason this guide exists is a study published in February 2025 in Transportation Research Part F, one of the main peer-reviewed journals of traffic psychology. It is open access, so you can read every word yourself from the sources section. Its charm is in the title, which quotes one of its own drivers mid-squeeze: "That's a bit of a tight squeeze!" Its value is in the method, because rather than asking drivers what they think they do in a survey, it watched what they actually did, and listened to them deciding in real time.

Peter Youssef, Ben Waterson and Katherine Plant of the University of Southampton's Transportation Research Group recruited nine drivers in spring 2022, each holding a full UK licence and each driving their own car, on a set route through Southampton chosen to serve up plenty of narrowings. A GoPro on the inside of the windscreen recorded the road and the driver's voice. The drivers were trained to think aloud as they drove, narrating what they were looking at and deciding. Afterwards, each driver sat down with the researchers and watched segments of their own footage back, explaining what they had been thinking at each narrowing, a technique called video-cued retrospective interview.

That produced roughly 45,000 words of verbalised decision-making across 175 narrow passage interactions, which the team coded using Neisser's Perceptual Cycle Model, a framework from cognitive psychology that traces how what you perceive updates what you expect, and how what you expect directs where you look next. The same framework has been used to analyse aviation incidents and, notably, the fatal 2016 Tesla Autopilot crash. A second, independent coder checked a tenth of the data, agreeing with the original coding 88.2% of the time (Cohen's kappa 0.84, which is strong agreement).

The first rule has official backing in the country where the study ran: the UK Highway Code, Rule 163, tells drivers to "give way to oncoming vehicles before passing parked vehicles or other obstructions on your side of the road". Irish instructors teach the same convention, and it is the basis of the meeting-traffic lesson in our own learner course. The second and third rules have no legal existence anywhere. They are what researchers call informal traffic rules, conventions held in common because everyone learned them by osmosis, and the study's drivers applied them as confidently as if they were statute.

There is a subtlety inside rule one worth teaching explicitly. The parked cars do two jobs at once: they decide the social question of who yields, and they create the physical question of where a yielding car can actually go. A driver on the blocked side is not just conceding priority, they are searching for somewhere to be: a gap between parked cars, a driveway mouth, a wide spot. Which leads to the study's most useful findings, because both of those jobs fail more often than you would think.

At narrowings blocked on one side only, where the unwritten rule gives a clear answer about who should wait, the study found that in 32% of interactions the driver on the blocked side went through first anyway. One time in three, the vehicle that "should" have yielded took the gap instead.

Before calling that bad manners, notice that the previous section already explained it. Sometimes the blocked driver has no gap to yield into, and as Section 6 shows, a driver with nowhere to go almost always goes through. Sometimes they are already committed by the time you appear. Sometimes the other driver waves them through. And sometimes, yes, it is simply someone chancing it. The reasons vary; the practical consequence does not: the rule tells you who is likely to yield, never who will.

The drivers in the study knew this perfectly well, and you could hear it in how they hedged. One expected oncoming drivers to stop when the parked cars were on the oncoming side, then added in the same breath that obviously not everyone does. Another summarised the working attitude: expect them to stop, but do not assume it, so slow down anyway. The researchers describe this recurring distrust and then, drily, note that it is not misplaced, because their own data shows the expectation failing a third of the time.

Start with how rare explicit signals are. Across 175 interactions, drivers flashed their headlights in 4.5% of them and used a hand gesture in 3%. The overwhelming majority of narrowings were negotiated without any deliberate signal at all: drivers read each other through movement. Speed, position in the lane, drifting towards a gap, holding steady. Pulling in was read as yielding. Staying out and maintaining speed was read as claiming the gap. One driver put the logic of standing still perfectly: the longer I stay, the more I am telling you to go.

Now the surprise. Received wisdom, and some of the research literature, says a flash or a wave at a narrowing means "I am giving way to you". But when the researchers checked who actually produced these signals, 85% of hand gestures and 47% of headlight flashes came from the driver who went through the narrowing first. The supposed signal of yielding was mostly coming from the drivers who did not yield.

The resolution is timing, and it is elegant. Every gesture and flash produced before the interaction was resolved, while both cars were still approaching and the outcome was open, did mean giving way, exactly as the textbooks say. The signals produced after the interaction was resolved, as one car was already passing through, were overwhelmingly a thank you from the driver who had been let through: 83% of post-resolution signals came from the car going first. Same wave, same flash, two opposite meanings, separated only by when in the encounter they happen. Drivers navigate this ambiguity so automatically that nobody had bothered to document it before.

Across the whole study, 91% of vehicles that had no suitable gap to pull into went through the narrowing first. Read that again, because it quietly outranks every other rule in this guide, including the priority conventions of Section 3. A driver with nowhere to go, goes. It does not matter much whose side the parked cars are on: if the oncoming driver has no space to yield into, the overwhelming likelihood is that they are coming through, and your decision is already made for you.

The skill this points to is one that experienced drivers do half-consciously and learners are almost never taught by name: scan for give-way gaps on both sides of the road, theirs as well as yours. The study's drivers can be heard doing exactly this, checking for gaps they could use, then checking what the oncoming car could use, and folding both into the decision. A German researcher, Arnulf Deppermann, had earlier proposed that drivers assess their own yielding spaces; this study's authors point out the data shows they assess their interaction partner's spaces too.

Vehicle type sharpens the same assessment. The study's drivers reasoned about what the oncoming vehicle physically was: a coach needs more room to pull in, a van cannot use a car-sized gap, and one driver watching a bus approach concluded they themselves were certain to be the one stopping, because the bus simply would not fit between the parked cars. The gap that matters is not the gap you can see, it is the gap relative to the vehicle that would have to use it. A space that yields easily to a Yaris does not exist for an articulated lorry, and the lorry driver knows it long before you do.

• Approaching any narrowing, ask three width questions in order: can we both fit through together? If not, where could I go? If nowhere, where could they go?
• If the answer to the last question is "nowhere", the interaction is resolved: you are yielding, whatever the parked cars say about priority. Find your space early and take it with decision.
• Size the gap for the vehicle, not for yourself. The bus, the bin lorry and the horsebox are all telling you the outcome from 200 metres away, if you think in their dimensions instead of yours.

1. Who has priority? Read the obstruction: their side, your side, or both? If one-sided, the unwritten rule gives an expectation. If both-sided, it goes to first come, first served, judged from who is nearer the narrowing.
2. What is the other driver's intention? Read their movement first: speed, lane position, drift towards a gap or towards the centre. Treat flashes and waves as secondary, and time-stamp them: before the outcome is settled, a signal usually offers the gap; after, it is usually thanks.
3. Where can either of us give way? Scan for gaps on both sides, and size each gap against the vehicle that would need it.
4. Can we both fit through together? Sometimes the narrowing allows a slow mutual squeeze. If it does, that changes the choice from "who waits" to "how slowly and how wide". If in doubt, it does not.
5. What about the vehicles behind the leader? If a convoy is coming, expect the informal convoy rule: the cars behind the lead vehicle will follow it through, and you will be expected to keep waiting. Count the queue before deciding your wait is over.

Notice what this list is: a structured observation routine for a situation most learners meet as an unstructured panic. It slots directly into the scanning habits from building observation skills and the clearance-versus-speed trade from our Roadcraft positioning guide, whose one-line rule applies to every narrowing: the narrower the gap, the slower the speed. If you cannot have the space, give up the speed.

The RSA's Making Your Mark marking guidelines contain a sentence written for exactly the streets this guide is about: "Where vehicles are parked or stopped on a narrow road, a fault should not be recorded for 'Clearance', provided the applicant passes by slowly, safely, and in full knowledge of the position of the vehicle." That is the whole bargain of the tight squeeze in one line. You are allowed to be close; you are not allowed to be quick and close, or close without knowing it.

Push the other way and the fault categories change. The React examples include an applicant who "drives into, or causes, or contributes unnecessarily to a 'bottleneck'", and one who is "driving towards parked vehicles on the left and does not move out in good time to pass them". Barging into a narrowing an oncoming car has already claimed is exactly the first of those; drifting late and abruptly around parked cars is the second. Deliberately restricting another road user's passage is marked under Right of Way. So the test punishes claiming a gap you should have conceded.

But it equally punishes refusing gaps that are yours. Progress is a fault category of its own, covering inability or unwillingness to make progress and causing obstruction, with unnecessary delay marked at the more serious grades. The candidate who stops for every oncoming car three streets away, or sits frozen at a narrowing they plainly own while a queue builds behind them, is collecting faults just as surely as the barger. What the tester wants to see is the study's Section 7 done visibly: read the obstruction, read the other car, decide, and then act on the decision without drama, at a speed that matches the clearance.

If a test result sheet has already told you this is your weak area, with marks under React, Clearance, Right of Way or Progress, our test result analyser maps each aspect on the RSA sheet to what actually went wrong, and the faults guide explains the grading.

This is Lesson 14 territory in our learner course, and it is one of the most coachable skills there is: a quiet residential loop at 5pm offers more narrow-passage practice in twenty minutes than a motorway offers in a year. If meeting traffic is the part of driving that still spikes your heart rate, that is a normal place to be, and it is fixable with structured practice.