How much space, and how fast, on a street where people walk

Two questions, usually asked in that order by anyone who drives through a town: how much space should I leave behind the car in front, and how fast is too fast where people are walking?

Both have published answers, from road authorities that answer them directly, in plain words. Neither has one answer. What follows is the shortest true response to each, then the qualifications, longest first — and the longest is that the authorities disagree.

The short answer

Leave at least two seconds, count them out against a fixed object at the roadside, and add seconds rather than subtracting them. Where people are walking, 30 km/h — 20 mph — is the number the public-health and road-design literature converges on, and several countries have now written it into law.

That is good enough to drive on today. The rest of this article is why it is not good enough to stop there.

The first qualification: the authorities do not agree on the number

The Highway Code says at least a two-second gap on high-speed roads and in tunnels where visibility is reduced, at least doubled on wet roads and up to ten times greater on icy roads, with the note that large vehicles and motorcycles need a greater distance to stop.

Spain’s traffic authority says two as well. Its own magazine puts the figure as the minimum needed between vehicles to avoid a rear-end collision, and gives the counting trick that makes it usable: pick a fixed point on the road, and when the car ahead reaches it say «mil ciento uno, mil ciento dos», because each of those Spanish thousands takes about a second to pronounce. It widens the gap to three or more seconds in bad weather or on wet asphalt, and to at least 100 metres or four seconds inside a tunnel when you are not overtaking.

Pennsylvania’s driver manual says four. Not two doubled for conditions — four, as the ordinary figure, with the same counting method: watch the rear bumper of the vehicle ahead, start counting when it passes a road marking or a roadside object, and if you pass the same object in less than four seconds, you are following too closely.

Virginia says two, three and four, and is the only one that explains what the disagreement is about. Its manual bands the gap by speed — two seconds under 35 mph, three from 35 to 45, four from 46 to 70 — and adds the sentence the others leave out: the space cushions do not work if you are speeding over 70.

Averaging these would destroy the only information in them. Read together, they say that a seconds rule is a rule of thumb whose correct value rises with speed, that two is a floor rather than a target, and that every one of them tells you to add seconds — for weather, for vehicles you cannot see past, for heavy traffic, for poor pavement, and for a car in poor condition. Take the largest figure that applies to your situation, and treat the smallest as the point at which you are already wrong.

The second qualification: what the seconds are standing in for

A gap measured in seconds is a way of holding a distance without arithmetic, and the arithmetic it replaces has two terms that behave differently. The stopping sight distance model used in road design writes it out: the distance is perception-reaction distance plus braking distance, computed as 1.47 × V × t + 1.075 × V² / a, with V the design speed in mph, t a brake reaction time of 2.5 seconds, and a a deceleration rate of 11.2 ft/s². The first term is linear in speed. The second carries the square. Double the speed and the thinking part of the distance doubles while the braking part quadruples.

That is why a time gap is the right instruction and a fixed number of car lengths is not: a constant gap in seconds scales the first term correctly on its own, at every speed. It is also why the number of seconds has to rise anyway, because the second term is not covered by the gap at all unless the car in front decelerates at the same rate as yours — and a car that hits something stationary decelerates a great deal faster than one that is merely braking.

The reaction figure itself is softer than it looks. Road design assumes 2.5 seconds because it is designing for a below-average driver. Virginia’s manual quotes hand response time as close to half a second and foot response as normally three-quarters of a second, then says in the next sentence that this excludes any delay in perception from a driver who is tired, on medication or distracted. The research summary that the section’s crash simulator cites for this constant gives roughly 0.7 seconds for an expected event, 1.25 for an unexpected one and 1.5 for a surprise, and its author states plainly that there can be no single number that applies universally. The spread between those figures is a large fraction of a two-second gap. That is the honest argument for four.

The second question: how fast, where people walk

Here the convergence is stronger than anywhere else in this article.

The World Health Organization’s speed management manual sets its limits by matching the speed to the worst crash the road can produce. Where crashes between cars and vulnerable road users are possible, the maximum it gives is 30 km/h; where the possible crash is side-on between cars, 50 km/h; where it is head-on between cars, 70 km/h. It supports 30 km/h zones for urban areas with regular pedestrian or cyclist activity, and reports that they lead to a reduction of more than 70 percent in serious injuries to pedestrians.

National instruments have followed, and they are worth reading as instruments rather than slogans. Spain amended its traffic regulation in 2020, in force from May 2021, to set 20 km/h on streets with a single surface shared by carriageway and pavement, 30 km/h on streets with one lane per direction, and 50 km/h only where there are two or more lanes per direction — a rule about street geometry, not about neighbourhoods. Wales moved its default limit on restricted roads to 20 mph in September 2023, and the Highway Code’s built-up-area rule now reads 30 mph generally and 20 mph in Wales. Virginia’s default for school, business and residential areas is 25 mph. British guidance on setting local limits adds a practical test: a 20 mph limit introduced through signing alone is likely to be complied with where the mean speed is already at or below 24 mph.

What the injury research says, and where it disagrees with itself

The curve everyone quotes is not one curve, and the two best-documented versions of it do not agree on level.

The American study behind the section’s simulator finds that the average risk of severe injury for a struck pedestrian — severe meaning an injury score of 4 or greater, deaths included — reaches 10 percent at an impact speed of 16 mph, 50 percent at 31 and 90 percent at 46. The risk of death reaches 10 percent at 23 mph and 50 percent at 42, and between 32 and 50 mph it rises by about 2.8 percentage points for each additional mile per hour, with a 95 percent confidence interval of 2.2 to 3.4. Two adjustments inside it matter more than the headline: the average risk of severe injury for a 70-year-old struck at any given speed is about the risk for a 30-year-old struck by a vehicle travelling 9.3 mph faster, interval 5.3 to 13.4, and the risk from a light truck is about the risk from a car travelling 6.3 mph faster, interval 2.1 to 10.6.

The German study, which the simulator keeps deliberately separate because it measures a different outcome on a different population, reports fatality risks substantially lower than those generally reported in the traffic safety literature, and explains the gap as sample bias: earlier work drew on crash samples weighted towards severe injury. Its conclusion keeps the shape while lowering the level — the fatality risk at 50 km/h is more than twice the risk at 40 km/h and more than five times the risk at 30 km/h — and it adds a point speed campaigns rarely make. Crashes above 50 km/h are sometimes judged virtually unsurvivable; on this curve they are not: at 75 km/h the estimated fatality risk is about 50 percent, with a confidence interval running from 26 to 68 percent.

British speed-limit guidance describes the same shape in a single sentence: the risk of a pedestrian dying in a collision with a car increases slowly up to an impact speed of around 30 mph, and above 30 mph the risk of death increases rapidly. The disagreement is about how high the curve sits, not about where it bends — and the bend is the part a driver can act on.

The section’s crash simulator, opened on a pedestrian at a crossing, is where the two halves of this article meet. Change only the approach speed and watch two outputs move together: where the car comes to rest, which is the quadratic term, and the speed it is still travelling at when it reaches the person, which is what the risk functions take as their input.

What we cannot tell you

We cannot tell you how many seconds is right for you, because the four authorities read for this article publish three different answers and each is correct about the roads it was written for. We cannot convert any of them into metres for your vehicle either: the braking half of that conversion depends on the surface, the tyres and the load, and the design model that produces such numbers says in its own text that longer distances should be used wherever practical and that it does not explicitly consider trucks.

We cannot give you a risk number for a person. Both injury studies publish population averages with confidence intervals, standardized to a particular vehicle fleet and population; the age adjustment alone shifts the equivalent speed by most of a 20 mph limit. A risk curve is an argument for a speed limit, not a prediction about the person in front of your car.

And whether following too closely, or exceeding a limit by a little, is by itself decisive when a claim is argued differs by market, and is stated in each market’s own traffic code and case law. What does not differ is the arithmetic. The space you left is the only part of the stopping distance you chose in advance.