Signals for Change: Ideas from the Dutch for Improving Traffic Lights

Dutch traffic signals are more efficient and safe for all road users than their Canadian and US counterparts. We have an opportunity to do the same here, and it starts with a change in policy.

I recently had the opportunity to work with Jason from the YouTube channel Not Just Bikes on a video comparing traffic signals in the Netherlands to those used in Canada. The popularity of the video and the discussions that followed demonstrate the very real desire to see improvements in Canada and the US. After all, intersections are where more than 50% of all urban traffic crashes occur.

Traffic signals in the Netherlands operate very differently than those in Canada and the US. They are designed to minimize delay for all users (and especially public transit), with advanced detection and adaptive operation. They also improve safety by separating more movements, like conflicts between right turning vehicles and pedestrians and cyclists.

Fully Protected Phasing

Traffic signals exist for the purpose of managing conflict where two streets intersect. But there are more conflicts than just the through-bound vehicles on each street – in fact, a four-leg signalized intersection has 32 conflict points! Left turning vehicles conflict with oncoming traffic, right turning vehicles conflict with pedestrians and cyclists. The conventional “green in one direction, then green in the other direction” does not attempt to manage these conflicts.

Conflict points at a typical four-leg intersection (Source: FHWA)

A more elaborate and safe intersection operation is to separate as many conflicts as possible with “protected phasing”. Protected phasing separates these conflicts in time – for example, by providing a right turn signal to separate right-turning vehicles from crossing the path of pedestrians and cyclists.

This Dutch junction has a “fully protected” right turn, meaning right turning vehicles are not permitted to turn when cyclists and pedestrians have a green across the conflicting path. Source: Reaperexpress

Adding these protected phases can add delays to all users at the intersection, but this can be mitigated by overlapping certain movements, like opposite left turns and right turns.

An illustration of right turns separated from bicycle and pedestrian movements by “overlapping” them with the left turns from the other direction. Source: MassDOT Separated Bike Lane Planning & Design Guide, Chapter 6

One of the downsides of protected phasing is that it increases the “cycle time” of an intersection (the time it takes for a signal to repeat all phases). Longer cycle times tend to increase delay for all road users, and can be especially cumbersome for pedestrians.

Pedestrian Half-Crossings

Half-crossings (technically called “two-stage pedestrian crossings”) are those where a pedestrian crossing is split into two separate signalized crossings with a centre refuge area. These are a hidden gem of large Dutch intersections. Historically, half-crossings are not seen as a pedestrian-friendly tool in Canada and the US, because they are usually not used in that way. The classic half-crossing is designed so that you cross the first half, then must wait for the entire signal to repeat itself before crossing the second half of the intersection.

A notorious half-crossing in Toronto. The crossing only provides enough time for pedestrians to get halfway, then they must wait a full cycle before crossing the rest of the way. Source: Google Maps

The Dutch use half-crossings in a way that doesn’t punish pedestrians, and they make intersections more efficient for all users at the same time. When a pedestrian pushes the button at one side of the intersection, they receive a “walk” signal for the first crossing. Meanwhile, the second half of the crossing continues to service traffic a bit longer and then changes to red, so that by the time the pedestrian finishes crossing the first half of the intersection, the “walk” signal is displayed for the second half. Pedestrians get the same level of service, and the unused intersection space is used more efficiently.

At very large intersections, this is also beneficial from an accessibility perspective. Pedestrians with mobility challenges are provided a refuge area halfway across the intersection, so that crossing a large intersection becomes less intimidating. Common accessibility guidance is to provide “rest areas” every 30 metres (100 feet) for someone to be able to stop. If the intersection crossing is more than 7 or 8 lanes wide, it’s likely the crossing is longer than 30 metres.

One of the biggest constraints to the operation of an intersection is accommodating the minimum pedestrian crossing time. At a very large intersection where a pedestrian must cross nine vehicle lanes, the pedestrian phase must be at least 30 seconds to give someone enough time to cross, even though the pedestrian occupies a very small area of that crossing at any given time. That becomes 30 seconds of intersection time that must be dedicated to the pedestrian crossing, which takes away opportunities for the intersection to respond to other needs such as transit priority.

At this Dutch junction, pedestrians cross a total of five traffic lanes over two separate crossings. Source: Google Maps

True Transit Priority

As proposed in the NJB video, when a bus or tram full of people approaches an intersection, it should be given priority on the basis of efficiency. If the bus is carrying 60 people, and the other vehicles are carrying one or two people, it just makes sense to prioritize the bus. The goal should be to reduce traffic signal delay for transit vehicles to as close to zero as possible. Such is the case at Dutch intersections; the bus/tram/streetcar approaches the intersection, and everything else stops to let it through.

This approach requires different technology but more importantly, a shift in mindset. Traffic signals are managed by traffic engineers. Traffic engineers (at least traditionally) are trained to manage vehicle traffic and maximize level of service for vehicles. Signals are carefully optimized to make sure every vehicle going through an intersection doesn’t get delayed too much.

Transit priority disrupts that mindset. It requires giving more attention to specific vehicles. The beneficiaries of this priority are transit agencies, and typically some level of service for vehicles must be “sacrificed” in order to make this happen. The profession in charge of managing the intersection is asked to accept a loss in performance for the benefit of something that (at least traditionally) they have no heightened concern for. Alas, we find a gap in policy.

Technologically, the Dutch prove that this is absolutely possible, indicating that we must be stuck at the gap in mindset here in Canada and the US. If strong policy is established to prioritize transit above all other intersection needs, then the technology solutions will follow. The City of Toronto made a positive step towards this recently when it shifted to optimizing “person-delay” instead of “vehicle-delay” when measuring intersection performance. The Toronto Transit Commission (TTC) has even worked with a major supplier to build a customized signal priority system, in response to policy demands and a lack of “off the shelf” solutions. Policy must lead, and technology will follow.

Flexible Signal Operation

The standard Canadian / US traffic signal is very predictable. Two to four “phases” combine to produce one signal cycle, and these simply repeat, in order, over and over. A phase might be skipped sometimes (such as when a left-turning vehicle is not detected) and cycles may change depending on the time of day, but other than that, you won’t see much change in the signal’s behaviour.

The Not Just Bikes video demonstrates that Dutch intersections are more efficient by responding in real time to the demands of the intersection. Rather than relying on the the “expected” traffic pattern at the intersection, Dutch intersections measure what’s going on now, by detecting incoming users in all directions. With that high degree of information, the signals can automatically optimize the signal performance by changing the order and length of signal phases.

Think about all of those times you’ve sat at a red light while the other direction remains green despite no vehicles being present. That’s a product of a signal that is timed for a “general” traffic condition, rather than one that is responsible to the live conditions. Flexible operation is not just better for cars, it’s better for pedestrians, cyclists, and transit users too.

Fewer Traffic Lights

Finally, one of the best ways to improve traffic lights is to actually have less of them. The theoretical capacity of a “free-flowing” vehicle lane is 1,800 cars per hour. Once you introduce signals though, that capacity is reduced based on the amount of green time that lane receives. If that vehicle lane now encounters a signal with timing split evenly between both directions, the capacity of that lane is reduced to less than 900 vehicles per hour (because of the time lost due to all directions receiving a red signal).

Now, to compensate for that reduced capacity, you need to add another vehicle lane. But adding another lane will make the intersection wider, which means the side street needs to receive more green time to allow pedestrians to cross the street, so more green time is lost on the main roadway. So another vehicle through lane is added to compensate, but this too results in less green time for the main street!

You can clearly see the diminishing returns.

In many cases it may be simpler to just remove the traffic light, reduce the number of travel lanes, and slow down motor vehicles to provide an environment where road users can negotiate with each other for space, rather than relying on a signal to set the rules for them. At high-volume intersections, a roundabout is likely the valid alternative. While Amsterdam received a lot of attention recently for turning off one of its traffic lights, most of the time the Dutch don’t even provide traffic signals to begin with. The Netherlands has 5,500 traffic signals in the entire country (0.4 signals per 1,000 people). By comparison, the island of Manhattan alone has 2,800 signals (1.7 signals per 1,000 people).

Overhauling the Canadian and US approach to traffic signals is not easy or simple, and will take time. More education will be needed. New technologies will need to be developed and installed. Standards will need to be rewritten. But again, we must ask the question: “if the Dutch can do it, why can’t we”?

It is clear that this is not an issue of an insurmountable technological hurdle. The hurdle we face is one of policy and practice. If we truly wish to prioritize safe and efficient travel for all road users, we must demand more of our traffic signals, starting with policy. If policy leads, technology will surely follow.

Special thanks to Reaperexpress for peer reviewing and assisting with the more technical content, especially surrounding transit priority.


  1. We’ve been adding adaptive signals along main corridors here in Pittsburgh (“Smart Spines” projects). Problem is they are bad with detecting bike or pedestrian needs; transit priority is possible but we need to install sensors in all our buses still and ensure they work well with signal sensors and controllers. We have very little room on our small streets for roundabouts or ideal TSP features like queue jumps. With how many ancient signals we have, it will be a while before we have widespread optimization. But we’re working on it 😉


  2. There are two ways which could potentially reduce collisions at intersections in the US that wouldn’t take much work but would be politically hard. The first would banish the right turn on red with the exceptions of very few places. If you are controlling the intersection, control the intersection. The second involves more work and cost. Put the traffic signals on the near side of the intersection for the people they control. This forces drivers, inconjunction with no right on red, to stop before the crosswalks and bike lanes. This also gives a drive who might not be paying close attention another car length to stop if needed because of inattention.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s