Jeff Speck: For a Walkable City, Remove Centerlines on Local Streets

Brighton Road, London, after the removal of the centerlines. Data from three roads studied suggested an average nominal speed reduction of 6.9 mph. Photo: Transport for London
Brighton Road, London, after the removal of the centerlines. Data from three roads studied suggested an average nominal speed reduction of 6.9 mph. Photo: Transport for London

Island Press has granted Streetsblog the exclusive right to publish a series of excerpts from Jeff Speck’s new book, Walkable City RulesThese will be shared over the next several months in conjunction with the book’s October release.

RULE 71: When repaving a two-lane, two-way street in an area where pedestrians are present, do not include a centerline without a site-specific justification.

Author Jeff Speck
Author Jeff Speck

Our friends at the Ministry of Silly Driving (a.k.a. Transport for London) had been suspicious for some time that streets might be safer without a center stripe. The city had already enacted in 2009 a “Better Streets” policy that embraced the Dutch concept of Naked Streets (see Rule 77), and engineers in the department were hopeful that a “less is more” approach might apply to centerlines as well. Holding to the initially counterintuitive logic of Naked Streets, these engineers expected that centerlines are one of many road markings that can make drivers feel more confident, causing them to speed as a result.

The city repaved three two-lane regional roads without centerlines, and compared driver speeds before and after. The results did not disappoint: when adjusted for the not insignificant impact of fresh pavement—which notably tends to encourage speeding—drivers on the reconfigured streets slowed down about 7 mph on average.

In the context of danger to people walking, 7 mph is a huge margin. At the speeds witnessed in this study, around 30 mph, a 7 mph reduction can cut the risk of death almost in half. As one considers driver behavior, this study’s outcomes are unsurprising. While making an effort to avoid conjecture, the authors suggest that some drivers “position their vehicles close to a white line regardless of the traffic conditions, believing it is their ‘right’ to be in that position.


Centerline removal introduces an element of uncertainty which is reflected in lower speeds.” They also note that the most conspicuous speed reductions occur when drivers see oncoming vehicles approaching.

It turns out that this study was not the first of its kind. The authors note an earlier effort undertaken by the Wilshire County Council between 2003 and 2007, which found that resurfacing streets without centerlines led to not only lower speeds, but also fewer injury crashes. And prior research by the UK’s independent Transport Research Laboratory had similar findings.

Based on all this evidence, with no opposing data, it seems safe to conclude that streets without centerlines are safer. Any public works department that insists on keeping them, without substantial evidence to the contrary, is likely valuing convention over human lives.

A few notes deserve elaboration. First, the increased speeds caused by resurfacing are real, averaging 4.5 mph in this study. The implied instruction worth sharing here is that, unless a safer striping configuration is being introduced, it likely hurts safety to resurface a road before mandated by deteriorating pavement.

Second, the study’s authors note that “not all roads would be suitable for removing central markings, particularly where the markings highlight a particular hazard.” There are exceptions to every rule, but purported exceptions must be reviewed critically.

Walkable City Rules, the upcoming book from Jeff Speck. Image: Island Press
Walkable City Rules, the upcoming book from Jeff Speck. Image: Island Press


211         Ryan Cooper and Sam Wright, “Centerline Removal Trial,” Outcomes Design Engineering, Transport for London (August 2014),

212         Ibid.

213         Ibid.

45 thoughts on Jeff Speck: For a Walkable City, Remove Centerlines on Local Streets

  1. This is really interesting. I would never think something as simple as removing a centerline could have this dramatic an effect of speeds. This is not only a low-cost tool, but it even saves money over the default of adding a centerline after a street is repaved.

  2. NYC streets would be safer if where possible one-way streets were converted to two-way streets, and the centerline was left off. Instead NYC converts two-way streets to one-way with re-designs and on street that get a road diet (4 to 3 lane conversion), a gigantic 10′ wide center area is created.

  3. i can confirm they work in my los angeles neighborhood.

    there are streets nearby my home that have street parking on both sides and no centerline, drivers approaching from opposite directions have to slow down because there are only inches for each driver to pass at the same time without hitting the parked cars or each other.

    usually, one driver will pull a bit over to give more space to the other driver.

    when there isn’t any other cars, they speed up.

  4. I lobbied hard to have the center line removed from a street close to our home. Thankfully the transportation department in our town was able to accommodate the request due to the road needing a slurry seal, and we haven’t had a center line for some years. I can confirm it makes the street immensely safer for cyclists and pedestrians as cars now give a lot more room and tend to slow down quite a bit, particularly when there is oncoming traffic. It does help that the roads near my home are quite about 19′. I would imagine it wouldn’t make as much of a different where roads are 25’+.

  5. File this under: If you want to do it better, don’t think too hard, just google ‘Netherlands street design’ and follow their lead.

  6. I think removal of centerlines is also a help to cyclists. When there’s a centerline, some drivers try to squeeze by the cyclist in order to not have to cross the line, or not cross it much. Without the centerline, they generally just go to the other side…

  7. SF striped bike lanes without a centerline on Kirkham Street to see what would happen. What happened, unfortunately, is that many drivers apparently felt uncomfortable with oncoming traffic and would drive with their two right wheels in the bike lane. When a centerline was added, motorists stayed out of the bike lane.

  8. That will happen in any event, with or without a center line.

    As a driver, if I am given a choice of being very close to a large oncoming vehicle with a combined velocity of (say) 60 mph and being very close to a slow moving, light vehicle (a bike) I am going to keep right every time. It’s not even something you think about – it’s just survival instinct.

  9. Research going back decades shows that total throughput is higher with alternating one-way streets over two-way streets. That is why almost every major city in the world has one-way streets

  10. I wonder if the efficacy of removing centerlines depends on traffic volumes.
    We have a low-volume 36′ wide residential street street (parking both sides) that was recently repaved, and we’re debating whether to reinstall the centerlines. Because it’s only used by about 1,000 cars a day, there is not much opposing traffic, so without striping, drivers have 22′ of street basically to themselves. I am thinking that in this case, installing a centerline will keep people closer to the parking lane, and traveling more slowly.
    I’d be interested to hear others’ thoughts or experiences.

  11. These findings contradict my own observations and research. I cycle daily on several 2 way streets without centerlines during my commute. These streets have known speeding issues (observed and supported by speed studies) and vehicles often drive in the middle of the street when there’s no oncoming traffic. Without centerlines, drivers often treat these streets like one ways with a single 22′ wide lane because of a lack of visual cues indicating otherwise.

    Without centerlines, the driver is dependent on oncoming traffic to indicate lane width. During gaps in on coming traffic, centerline-less streets visually indicate a very wide street which leads to excessive speeds.

    I’d be very careful about putting Speck’s recommendations into practice without additional research.

  12. “That will happen in any event, with or without a center line.”

    That’s incorrect in this example. Did you travel down Kirkham Street before and after the centerline was put in? After the centerline went in, drivers stayed out of the bike lane. Reasoning is pretty straightforward – you can see oncoming traffic is on their side of the stripe while you are on your side, so you are willing to tolerate a closer pass.

  13. Thank you for that information. But that raises the question, does the city want to maximize for higher throughput? Or maximize for safety which means lower throughput?

  14. I was thinking this could be the case. It makes sense that average speeds would go down but would the outlier top speeds increase?

  15. As Mark Brown above pointed out, that’s a real possibility but I’m sure it’s a case by case basis. In his example the travel lanes were 11′ but yours sounds like it’s 10′ so it might not be as much of a problem. It would be great if you could measure speed without the lines and then install temporary lines and re-measure.

  16. Indeed. It would be handy to know if there’s a volume/width rule of thumb.
    Well, the lane width depends on how wide we want to call the parking lane (7 or 8 feet). In practice it’s unlikely most vehicles will be more than 7′ from the curb, so the lane is effectively 11′, perhaps slightly wider.

    If we do install centerlines we’ll be sure to measure speeds. We already have the before speeds (28 on average with an 85% of 33, speed limit 25)

  17. That agrees with my observations. I often see drivers who seem more focused on staying to the right of the center line than keeping a safe margin to cyclists on the right. Remove the center line and drivers have one less thing to worry about, freeing up attention to keeping a good margin on the right.

  18. This has been observed and demonstrated elsewhere, and the missing issue in the discussion is context. Parking/no parking, lane width, volume, etc all have bearing on the issue. I’ve had people insist that one street is wider than another and is the reason for the speed disparity, when in fact the lane lines made the street appear narrower, thus slowing traffic. Each case is different.

    Speck is notorious for throwing out one-size-fits-all solutions. He claims one way streets are inherently more dangerous without any data, and often in contradiction of the data. Again, it is contextual.

  19. In that shared situation the bicyclist should control the lane, especially in an urban environment. Unfortunately most cyclists are scared to do so, inviting the close pass.

  20. The Dutch don’t routinely omit the centerline. As noted above, it isn’t a single approach. The Dutch naked streets are typically in dense urban areas where the priority is more in favor of bicyclists and pedestrians. They may omit a centerline on a more rural connector road, but those are in “walkable” areas as noted by Speck. This story omits the crucial contextual issues.

  21. Two way streets have more conflicting movements at intersection and thus are often more dangerous. One way streets don’t necessarily result in higher speeds as is often claimed. The data speaks truth to that and the one-way/two-way canard has worn thin. Speck flogs that dead horse despite frequent evidence to the contrary.

    Left-turning traffic on a two way street is the movement most associated with ped crashes at intersection. There are certain crash typologies for all modes (biker/ped/driver) that are eliminated with one-way streets.

  22. Hi Mike. Yeah if they are actually striping a bike lane that may be different My experience here is just with and without the centerline, when there is no bike lane .

  23. What really annoys me is when I’m riding on a road and there’s clearly no oncoming traffic for a few hundred feet or more, and yet a car behind me honks when I’m taking the lane, like they are somehow incapable of crossing that centerline..

  24. As freeways and highways jam up, more drivers are taking two-way local streets as bypasses. This includes residential streets! Very bad trend! SOLUTION (perhaps better than eliminating yellow line.): Every third or fourth intersection, ALLOW ONLY BIKES TO GO THRU. Force drivers to turn right. No left turns allowed. EFFECT: Impatient commuters will totally avoid that street. Local drivers will use the street, SLOWLY, to get to their homes. Vancouver, B.C. does this. Very effective!

  25. Hmmm…. sorry, but you are misinformed. Please hop on ‘maps’ so you can get a better idea in general about NL street design and engineering. Paradoxically, the place you see a center line is on their separated bike infra.

    About your other point. I agree in theory. However if we are speaking about the United States where so many policy makers & government employees engage in expensive studies to figure out how to make our cities better, then simply ‘looking to the Netherlands’ is absolutely appropriate and a much better use of their time.

    And here the crux: NL residents drive exponentially fewer miles in cars than us Americans. It’s chicken – egg but you can’t get NL results with U.S. driving rates. The U.S. is currently attempting to improve neither….

  26. Double yellow lines technically mean you cannot cross them even if there is an obstruction in your way.

  27. I generally assume 8′ for parking lanes but if you want to go with 7′ that works too, in which case the other travel lanes are 12′ so people would definitely be speeding.

  28. We’re only at 36’ wide so 11’ lanes but yeah
    (also we have a number of 6’ parking lanes. Narrow streets in pittsburgh)

  29. In all cities I know of the traffic/transportation department has multiple goals including both throughput and safety. Inevitably there is a trade-off and compromise between those goals. The voters do not support 100% safety if it means they can’t get anywhere efficiently.

    So even though 2-way streets may statistically be safer that does not mean the voters will support them.

  30. I don’t know anything about Kirkham Street but I do know that I prefer driving further away from oncoming traffic. Whether there is a line or not I will usually keep as far left as I can.

    There may be drivers who will wander over to the left if there is no center line but I drive with more discipline than that.

  31. A cyclist who rides further to the left places himself closer to oncoming traffic and even closer to traffic going in the same direction who have to pass the cyclist.

    You can get away with that if and only if you can ride at the prevailing speed of traffic, but otherwise you will be constituting a hazard to drivers and putting yourself at extra risk

  32. In PA you’re permitted to cross the double yellow to keep a safe (4′) passing distance next to cyclists (assuming it’s safe to do so)

  33. A residential street can become like a racetrack. The solution is other traffic calming measures like speed bumps.

  34. Ah, but we have slightly over an 8% grade on the street in question, which is just above the PennDOT recommended limit for using speed humps ¯_(?)_/¯

  35. That’s a good point. There are many variables to consider. I think that if the centerline caused drivers to see a narrower lane, they might slow down. But if they see two wide lanes, especially with a white lane along the side, they might feel they can just speed through that lane and pay less attention to pedestrians.

    But if removing road markings helps people think of the road as a shared space, that might help with speed and pedestrians. There is not a one size fits all solution.

  36. Could there be a speed “divot?” The grade can change from 8% to something less, then back to 8%. It would probably cause people to bottom out. But maybe that’s a good thing.

  37. Interesting about Kirkham St. Looking at GMaps, I can see how without the centerline drivers would veer into the bike lane when an oncoming car is approaching. Does this change for roads without bike lanes?

    I noticed that this week SF added a yellow centerline to Fillmore St b/w Haight and Duboce, as part of a repaving. It’s a 2-lane road (1 lane in each direction), with car parking on both sides. I don’t know yet what the effects will be – I’ve only commuted on it for a few days this way, but am considering emailing SFDPW to ask about it.

  38. As an isolated strategy, center line removal can cause vehicular speeds to increase for the the reasons that some of the comments have put forth. If one has good sight distance and an expectation of no conflicts, a driver can easily drive faster without feeling they’re doing something dangerous. These particular studies show a decrease in speed (as do most, but not all, studies of centerline removal). The actual effect is very dependent on the physical setting.
    One excellent way to keep vehicle speeds down is to install advisory bike lanes (ABLs) with narrow center lanes. The picture shown at the top of this article is of a British ABL with a wide center lane. Narrow center lanes tend to keep vehicular speeds lower. Unfortunately, Britain tends to use ABLs with wide center lanes which tends to keep both volumes and speeds high. And drastically reduces safety and comfort on these streets.
    The crash modification factor (CMF) for centerline removal is very close to 1 which means that research in the US has found that strategy to have little impact on safety. As with all CMFs, the research is only applicable to the types of roads and environments the study covered.

Leave a Reply

Your email address will not be published. Required fields are marked *