Wednesday 25 April 2018

Safe roundabouts revisited: There's increasing evidence that one particular design really is the safest for cyclists

The safe roundabout design for cyclists looks like this. Please
read my previous blog post and watch the accompanying
video as both of those describe the features of this design
There's been quite a lot of news recently about roundabouts in the Netherlands and I'm pleased to say that the new statistics are highly supportive of my previous article about how a truly safe roundabout should be designed for cyclists. Here in Assen we use this safe design instead of the design used in much of the rest of the country and as a result far fewer cyclists are injured on our roundabouts. Injuries occur elsewhere quite frequently, and it estimated that between 100 and 150 more cyclists a year end up hospitalized because of the unsafe design adopted elsewhere, but the safe design greatly reduces their rate to the point that I don't know of any which have occurred on the particularly safely designed roundabouts in this city. Please go and read my previous article linked above to find out safe roundabout designs.

Now on with the new content:

Dangerous roundabout designs are a continuing issue in the Netherlands
Data from the Smart Traffic Accident Reporting system has been used to create a map which shows the most dangerous roundabouts in the Netherlands. I've picked out below some of the Dutch cities which I've written about before and shown how they rate for roundabout safety. There's a colour code in use. Roundabouts which have caused no problems are not shown, roundabouts with a yellow dot have seen a small number of collisions, orange dots are worse and red dots are the worst locations of all. The colour code doesn't tell whether collisions resulted in injuries.

The outcome is obvious. Every city in the Netherlands which uses the dangerous cycle priority roundabout design, regardless of the details of how their roundabouts are built, has a heightened rate of injury of cyclists. To give a point of comparison, first consider Assen, which has only the safe roundabout design:

The interesting thing about the map is that it so strongly reinforces the data which I based my previous blog post upon: Only seven of Assen's 21 roundabouts are visible on this map because almost no crashes every occur here on the roundabouts due to the safe design. On those rare occasions when there is a crash it usually results only in minor bodywork damage to cars and that is the case for all the crashes which occurred. Between 2014 and 2017 there were just eight minor crashes with material damage to cars between all 21 roundabouts in Assen. There were no injuries. Annual crash rate: 1 for every 32000 population per year. Injury rate: 0. No-one at all was injured on a roundabout in Assen during this period of time.

Groningen is our first example of a city which gives cyclists priority on the roundabouts. Here we will not see the same zero injuries result. Note that on this map we have not only yellow dots but also orange and even red dots indicating sites with far more crashes. Not only were there considerably more crashes, 96 in total, but 26 people were injured severely enough that they had to attend hospital as an in-patient as a result of crashes at Groningen's roundabouts in the same 2014-2017 period. Groningen's population is three times greater than Assen but clearly the roundabouts in that city create far more than three times the danger. Crash rate: 1:7500, Injury rate: 1:28000. i.e. one in twenty-eight thousand of Groningen's population is hospitalized each year by the roundabouts in use in that city.


Zwolle's population is 125000 which makes just under twice the size of Assen. This city also uses the dangerous design of roundabout and again you see orange and red dots where many crashes have occurred. 105 crashes resulted in 37 people being hospitalized at the roundabouts in Zwolle between 2014 and 2017. Three years ago I identified one of Zwolle's roundabouts as the most dangerous place in the city for cyclists and it's highlighted on the map above having caused three more injuries. Crash rate: 1:4800, injury rate: 1:13500. One in every 13500 people in Zwolle is hospitalized each year by Zwolle's roundabouts.

Please read a more recent update about the danger of the roundabouts in Zwolle. Zwolle went to the effort of rebuilding roundabouts but they chose a different unsafe design to replace the unsafe design they already had. As a result, even the local government of the city has had to admit that three of the top ten most dangerous places in the city for cyclists are roundabouts.


The population of 's-Hertogenbosch's 152000, which makes it about two and half times the size of Assen. The city uses the dangerous design of roundabout and they caused 105 crashes with 34 injuries over the 2014-2017 period. Three years ago I assessed two of the two of Den Bosch's roundabotus as being the most dangerous places for cyclists in the whole city and one of those is highlighted above showing the that it has caused three more injuries. Crash rate: 1:5800, Injury rate: 1:17900. One in every 17900 people is hospitalized each year in Den Bosch each year due to the roundabout designs used.

Enschede's population is about 160000 which makes it also about 2 and half times the size of Assen. Enschede was the first city to use the dangerous "priority" roundabout design. Three years ago I identified the roundabout highlighted above as the most dangerous in the city for cyclists and clearly it has continued to injure people. In total, 80 crashes occurred at Enschede's roundabouts between 2014 and 2017 and 10 people were wounded. Crash rate: 1:8000, Injury rate: 1:64000

The same pattern is seen even in small cities which use the dangerous roundabout design. Middelburg is a small city in the South West of the Netherlands with a population of just 48000, about 2/3rds the size of Assen. There are relatively few roundabouts but the design used is the dangerous "priority" design. Middelburg's roundabouts have caused a total of nine crashes in which five people have been hospitalized. A crash rate of 1:21000 and a hospitalization rate of 1:38000 per year.

Drachten is also smaller than Assen with about three quarters of the population yet its roundabouts have caused a total of 32 crashes, 8 injuries and one death over the period when Assen's roundabouts caused no injuries at all and just eight minor crashes. The roundabout which is highlighted for Drachten is the shared-space "squareabout" at the Laweiplein, which on its own is responsible for five crashes and two injuries. Unfortunately, there are many exaggerated claims made for the safety of this junction when it is actually a very unsafe design. Please read my previous post on the subject of the Laweiplein for more details. Crash rate: 1:5600, Injury rate: 1:22500, Death rate: 1:180000

More Shared Space roundabouts ?
Given the poor track record of Shared Space in general and the poor track record of the Shared Space roundabout in Drachten, you might think people would steer clear of the concept. But that's not happening: A new shared space roundabout was recently built in Winschoten and it's got such a bad record already that it was recently recognized as the most dangerous roundabout in the entire country with just one location having caused 13 crashes and 3 injuries. Winschoten's population is only about 18000 but this roundabout is so bad on its own that we get a crash rate of 1:5500 and an injury rate of 1:24000 per year for the population while ignoring everything else in Winschoten.

Winschoten's Shared Space roundabout - recently chosen as the most dangerous in the country. Truly a horrible design. No-one knows what they're supposed to be doing and uncertainty results in accidental risky behaviour.

Are we designing for efficiency and safety ?
The crash and injury rates quoted for each city above are a primitive way of making a comparison, simply dividing the population by the average annual rates of crashes and injuries without any other factors, but nevertheless, the point as does all the other data, towards Assen's roundabouts being amongst the very safest in the Netherlands. So why isn't the rest of the country copying this safe design ?

We shouldn't be surprised by this result. It was predicted 13 years ago.
It's been known for many years that the design of roundabouts which gives cyclists priority over motor vehicles around the edge of the roundabout, relying upon drivers always to spot a cyclist who might emerge quickly from a blind spot and assuming that all drivers pay attention to ensure the safety of others as they drive, inevitably results in more injuries. In 2005 the extra injuries resulting from this design were estimated as 52-73 people requiring hospitalization (in-patient treatment) each year.

Roundabouts in the Netherlands over time
Since 2005, the number of roundabouts in this country has nearly doubled. It is also claimed that more journeys are being made by bicycle than before and there are more cars. Therefore it's reasonable to assume that the number of extra injuries requiring hospital treatment due to the priority roundabout is now around 100-150 per year. That is perhaps why the injuries on roundabouts are now getting more press attention than before. But these extra injuries shouldn't be a surprise to anyone because this was predicted before half of the roundabouts which exist now even existed.

If more than a hundred people per year are being injured due to a choice of roundabout design when a much safer design which is proven and already in wide use could be adopted, there needs to be some other justification for the unsafe design. Typically, people claim that it's more efficient for cyclists to have priority, but unfortunately the "priority" is merely part of the name of the design and not any proven effect. In reality you can't ride around a priority roundabout as quickly as you can ride over the safe design because you have to be quite cautious when you're relying on others always to look out for you. In fact, many "priority" roundabouts are quite unpleasant and inefficient for cyclists because we are forced to cross more lanes, which means slowing considerably and repeatedly checking whether drivers approaching in both directions who are supposed to have noticed us actually will give way to us, and the much tighter corners on the priority design often cause trouble.

Are we just adopting the "priority" design because that word sounds good ?
A few days ago I rode through a particularly poor recently built example of a "priority" roundabout, one which bizarrely has been held up as a good example locally. This morning I went back to the location especially to make a video showing how extraordinarily inefficient a "priority" roundabout can be, and how it can expose the cyclist to needless extra danger:

This video shows my pick for the most inconvenient newly built roundabout anywhere in the North of the Netherlands

Not all of the poor features of this roundabout and surrounding infrastructure are included in the video. This gentleman clearly had ridden on the "wrong" side of the road, treating a narrow unidirectional cycle-path as bidirectional. People do that if the infrastructure makes doing the wrong thing more convenient than doing the right thing. Once you reach a priority roundabout when travelling in the wrong direction, all bets are off for safety.

If I had made a return trip through the roundabout in the video I would have ridden here. This skips the indirection to the other side of the road, but instead riders have to cross three lanes of traffic, the first of which is a bus lane, and pass over two islands, neither of which is wide enough to offer safety even to a child's bike. I suspect that in reality few people go all the way around in the video so there will be many people riding in the wrong direction here. That is far more dangerous because they'll be relying upon drivers looking in both directions at once across three lanes for their safety and they haven't got any refuge to hide in.

Stills from the video above. How can anyone in any seriousness claim that the "priority" roundabout on the left, which requires cyclists travelling South to North to turn sharply and cross six lanes of traffic to go all the way around the roundabout, is safer or more convenient than the one on the right where cyclists cross two lanes of traffic in a straight line with good sight-lines ?
The safe design is well established and has a proven track record. It also is naturally resistant to error in implementation due to imperfect copying of the design, making an ideal basis for other countries to create their own safe roundabout designs for cyclists.

The "priority" design is proven to cause a significant level of injury to cyclists in the Netherlands and even here is misinterpreted quite often in such a way that it is more dangerous than it need be and causes significant inconvenience to cyclists, as shown in the video above.

The roundabout design which should be copied by countries outside the Netherlands is therefore the safe roundabout design. Copy what is proven to work, not merely anything that is "Dutch".

On our study tours we demonstrate the difference that good infrastructure makes. Roundabout designs are included.

Danish Study
A new study from Denmark provides more supportive evidence. In particular (from the Abstract): "Single-lane roundabouts with separate cycle paths, where cyclists must yield to motorists entering or exiting the roundabout, are safer than roundabouts with cycle lanes." It doesn't matter which country they are inplemented in: the principles of safe roundabout design remain the same.

Update December 2019 - new Dutch study
A new Dutch study which analyzed the safety of all roundabouts in the Netherlands has shown that four times as many cyclists are injured on roundabouts where cyclists have priority as on those where they do not. The study further notes that roundabouts with bidirectional cycle-paths are safer than those with unidirectional cycle-paths. It is of course the case that almost all of the safe without priority roundabouts use bidirectional cycle-paths while these are less common (and less safe) on the with priority roundabouts, so that's further supporting evidence


Wardski said...

I have to admit, however, that even after scanning your May 24 2014 article on safe roundabouts, I am not totally clear on the main points that make it exactly clear. (I remember reading the article before, that's why I just skimmed it now).
Am I right that the main points for a safe roundabout are:
1) "The design of Assen's roundabouts results in crossings of roads always being at 90 degrees and the 90 degree crossing point being where cyclists must give way."
2) No priority for cyclists.


David Hembrow said...

That's part of it. Here's a longer list:

1. 90 degree crossings to improve sight-lines of cyclists.
2. Not relying upon drivers to take decisions which maintain the safety of cyclists but giving that control to cyclists themselves.
3. Adverse camber for cars going around the roundabout to slow cars down.
4. Refuges between streams of motor traffic which are wide enough to accommodate a whole bicycle.
5. Bidirectional cycle-paths because they allow cyclists to cross fewer streams of traffic (crossings are where the dangerous interactions occur).
6. Right turns with no interaction at all with motor vehicles.
7. Layout which makes it very obvious to everyone what they should do.

Watch the video which accompanied the older blog post as that demonstrates these features.

hardcoder said...

Thanks for this thoughtful post. The Assen roundabout is very large, do you think space constraints ever determine the choice between these designs? (By the pics/vids the other ones don't seem significantly different in size to me.) Do you know of examples in a smaller and/or urban context?

We have a similar problem on a Melbourne intersection at the moment ( It's a busy north/south cycle route, and a east/west rat-run for cars. Cyclists have priority, but are getting cleaned up by motorists failing to give way. I'm loathe to yield priority to crossing motor traffic, but the stats are terrible (some serious injuries occurring). Right now the council responsible is looking for ways to reduce traffic flow, or provide some intervention that works. They have implemented a "c-roundabout" elsewhere which has apparently worked well in context and hugely improved safety ( The roundabout imposes a larger horizontal deflection on entry, and cyclists do not automatically have priority.

Also, while most Dutch towns have me sighing longingly, that town square in Winschoten looks pretty lousy. It looks more like a London 'shared space' project :) Just kick the cars out of the town square!

David Hembrow said...

Hardcoder: Thanks for your comment. I'm not sure if you've seen all 21 of Assen's roundabouts (you can find them here). All of them, including the roundabout highlighted above and in 2014, are in an urban context, within the city. I choose that one to highlight because it is clearest to understand photos of, but the others are pretty good too and some of them are a bit smaller. If there's not room for a safe roundabout to be built then we shouldn't build an unsafe roundabout as a compromise, but something else altogether. Perhaps remove motor traffic to some other place or use traffic lights.

Your Melbourne road is an interesting example. It's looks like you have plenty of space. There are very wide car lanes, car parking on both sides and also palm trees on a wide green in the middle, yet the only provision for cyclists is a dangerous door zone cycle-lane. This could all be re-arranged to make life much better for cyclists and not do any harm at all to drivers. Take a look at the layout of this street in Assen which at 18 m width (including everything up to the property owners land on both sides) I think is narrower overall yet provides better for everyone.

The C-roundabout is interesting, but I'm not sure what the advantage is supposed to be. The deflection of traffic will make motor vehicles come more closely to cyclists, and cyclists are completely abandoned at the the roundabout. No infrastructure to help at all. This seems a missed opportunity.

From the point of view of a cyclist, statistics around roundabouts require some analysis. It's often quoted that "roundabouts are safer" than one junction type or another, but it's usually the case that the comparison is between a roundabout and an unsignalled crossing (people often erroneously think roundabouts are safer than traffic lights but I've not seen evidence to support this) and the increase in safety observed is usually safety for car occupants and not for vulnerable road users like cyclists and pedestrians who in many cases are worse off due to roundabouts. The safe roundabout for cyclists which I have written about above and elsewhere is quite unusual in that it is genuinely safer for cyclists than an un-signalled crossing. In Dutch cities which have adopted safe roundabout and traffic light designs, it is the minor un-signalled junctions left over which stand out as the least safe places for cycling.

BTW, this traffic light junction on the same 18 m wide street is extremely safe and convenient for cyclists and I think this would also fit easily on your street.

Unknown said...

Hey David, hope you're fine. Just like to congrats you, your blog is really great. I'm doing a post graduated course about Urban Mobility and as a semester final study case project i chose Groningen and your blog was really helpful to me.

Wish u the best.

zerakith said...

Any idea on flow rates through roundabout? It seems like quite a quiet junction for cars. Does it need reduced car use to sustain this design?

David Hembrow said...

zerakith: The recommendation in the Netherlands is that roundabouts also used by cyclists should have no more than a maximum of 500-1500 motor vehicle movements in any single hour across any of the arms of the roundabout. See a previous blog post which discusses this. In other countries, such as the UK, roundabouts are used with far higher motor vehicle flows, but these are never safe for cycling.

If you have greater traffic flows then grade separation or traffic lights should be used if there are cyclists in the area, or turbo roundabouts can be used if there are absolutely no cyclists nearby (e.g. motorway junctions).

Anonymous said...

Hi David and thank you for your informative blog. I noticed in the Nobellaan-Thorbeckelaan example that three arms have zebra crossings but one does not (which is the closed to a school too) and wondering why this is the case? Writing from Melbourne, where guidance says to be consistent on all arms (although somewhat haphazard implementation of this and other design components to date eg ).

David Hembrow said...

Hi Unknown,

You're right about the Nobellaan roundabout. One of the arms does not have a zebra crossing. It does still have a separate crossing for pedestrians but for some reason this arm does not indicate to drivers that pedestrians should have priority. Why ? I have no idea. No explanation at all for this. To me it seems like an oversight.

The Melbourne example is interesting to see. It's far from the worst example. The waiting areas between lanes appear to be generous and it would appear that the sightlines are good. Just two things to note really: 1. I hope they're not attempting to give cyclists priority as this is especially dangerous with a bidirectional cycle-path. 2. It's unfortunate that the cycle-paths are only on one side of the road as bidirectional paths on both sides give cyclists a lot more utility.