Evidence That Split-Phase Signals Are Safer Than Mixing Zones for Bike Lanes

Mixing zones, rendered above, are DOT's standard treatment for left-turns on corridors with protected bike lanes. Image: DOT
Mixing zones are DOT’s standard treatment for intersections where motor vehicle traffic turns across the path of protected bike lanes, but they are not as safe as intersections where pedestrians and cyclists get exclusive signal time. Image: DOT

When DOT presented plans for a protected bike lane on Sixth Avenue, one point of contention was the design of intersections. How many intersections will get split-phase signals, where cyclists and pedestrians crossing the street get a separate signal phase than turning drivers? And how many will get “mixing zones,” where pedestrians and cyclists negotiate the same space as turning drivers simultaneously?

DOT tends to opt for split-phase signals only at major intersections, like where Sixth Avenue crosses 14th Street and 23rd Street. At other cross streets with turning conflicts, the mixing zone is the go-to treatment. Manhattan Community Board 4 wants to change that, asking DOT to include more intersections with dedicated crossing time for pedestrians and cyclists in the Sixth Avenue project.

The evidence backs up CB 4’s assertion that split-phase signals are safer. Data from previous protected bike lane projects in Manhattan show that the reduction in injuries on streets that mostly received split-phase treatments was more than double the improvement on streets that mostly received mixing zones.

A 2014 DOT report [PDF] analyzed three years of before and after crash data from Manhattan’s protected bike lanes. The last section of the report shows the change in total crashes with injuries on 12 protected bike lane projects — six with primarily split-phase treatments (segments of Eighth Avenue and Ninth Avenue below 23rd Street, and two unconnected segments of Broadway in Midtown), and six with primarily mixing zones (segments of First Avenue, Second Avenue, Columbus Avenue, Broadway, and Eighth Avenue above 23rd Street). We don’t have access to the raw numbers DOT worked with, but the aggregate data strongly suggests that split phase treatments are significantly safer.

On average, crashes with injuries declined 30 percent on the six “split-phase” redesigns and 13 percent on the six “mixing zone” redesigns.

On eight protected bike lane segments, DOT also analyzed cyclist injury risk, a measure that, roughly speaking, divides the number of cyclists killed or severely injured on a given street segment by the number of cyclists counted on that street. Two of those segments, on Ninth Avenue and Broadway, were primarily split-phase projects while the others were primarily mixing zone projects. While we don’t have access to DOT’s raw data, the average improvement on the split-phase segments was about 51 percent, compared to a 30 percent improvement on the mixing zone segments:

Image: DOT
Image: DOT

Mixing zones cost less than split-phase intersections and require removing fewer parking spaces, which means they can be implemented faster and with less pushback than split-phase intersections. In November, DOT officials also told CB 4 that they were wary that, in cutting pedestrian crossing time, split-phasing would lead to increased disregard for walk signals.

Last Wednesday, DOT came back to CB 4 without having added any split-phase intersections to the Sixth Avenue plan. The committee passed a resolution calling on DOT to install two additional split-phase signals on Sixth Avenue in the next year, and DOT’s Ted Wright said the department could study their feasibility in the future.

The data suggest that pedestrians, cyclists, and motor vehicle occupants all get a clear safety benefit from split-phase signals, however. Even if split-phase signals aren’t included in the first iteration of a project to ensure timely implementation, it should be standard practice to retrofit existing protected bike lanes with them over time. More injuries and deaths will be prevented, and more people will feel safe biking on city streets.

  • Reader

    “Mixing zones cost less than split-phase intersections and require removing fewer parking spaces, which means they can be implemented faster and with less pushback than split-phase intersections.”

    Cost, parking, and time. Three things that shouldn’t be weighted so heavily by the city if it’s serious about Vision Zero.

  • Joe R.

    The problem is split phase signals usurp the cyclist’s green time, which means they likely won’t be obeyed, resulting in the same conflicts. Easier solution is to just ban turns on the side of the street with the bike lane.

  • Geck

    My personal experience is that mixing zones do create a stress zone in otherwise lower stress protected bike lanes, but that many cyclists and pedestrians do ignore the split phase signal and proceed, with parallel traffic, through the intersection against the light.

  • Simon Phearson

    Yes, this. I usually ride in bike lanes and observe traffic signals, but regular split-phasing would break that habit. A split-phase on a protected lane may be “safer,” but it ensures that, as a cyclist, you’re standing around a lot longer than any other traffic, because pedestrians will jaywalk when it’s clear and non-turning traffic will have the green throughout your stop-phase.

    Please, please, please – change the law so that we don’t have to use this infrastructure if we don’t want to.

  • BBnet3000

    While I agree that the number of turns should across the bike lane should be reduced, I don’t see it as a huge problem if people roll through the light. Most people will do this when no car is turning, ie. when there is no conflict.

  • BBnet3000

    These don’t just cut down on safety, they also cut down on convenience (a forbidden word I know when it comes to cycling in New York).

    There are a number of busy turn corners where a car turning will frequently stop sideways across the bike lane and the cars lining up behind form a tight line, pretty much blocking all through movement for people cycling with the green light.

  • J

    Yes, this is a HUGE problem. DOT isn’t really interested in trying new ideas, like protected intersections.

  • Joe R.

    I don’t see it as a problem either but I can see these mixing zones becoming yet another ticket trap for the NYPD. The red should be changed to flashing yellow so that a cyclist is not legally required to wait when no cars are turning. Or you could put sensors so the light only goes red when cars are turning but of course that’s too sophisticated for NYC. Heaven forbid we actually tried to have traffic signals which only go red if something is actually crossing.

  • Joe R.

    Note that where those are used overseas, you generally only have one such intersection per km or so. In Manhattan you might use protected intersections where the avenues intersection major cross streets. Using them every 250′ would be ridiculous. And I don’t think the geometry would work well with a narrow 30′ cross street.

  • every single bike rider here does what i do–you go around the car that t-bones the intersection by failing-to-yield (haha i heard a rumor that’s illegal but i’ve NEVER seen it enforced even when i’ve stopped and walked a cop over to show her the infraction as it was happening!). in going around that car, you are of course heading into moving traffic with VERY tight space to make the maneuver, and then you are heading back into the bike lane with jaywalking pedestrians having taken advantage of the bike lane being blocked.

    so the scoreboard: the cops won’t protect us, the cars don’t care, and pedestrians make it more dangerous for us.

    please someone lecture me about how cyclists have to follow all the rules again!

  • chekpeds

    Exactly – as a pedestrian you are scared, as a cyclists you have to put the foot down every 200 ft …

  • chekpeds

    Thank you David . split phase signals are better for all. in our experience DOT may come back in 5 or 8 years and it takes an enormous fight.. These features should be part of the basic package, at least at dangerous intersections. IN the case of 6th AVenue 66% of the injuries will not be addressed because of the last of split phase.. How are we ever going to reach vision Zero?

    Do you buy an unsafe car because you can get it next month , instead of waiting six months and get a safe car? a bike lane is the same.

  • You don’t need them everywhere, but at more complicated crossings they’d help. I always think Jay/Tillary in Brooklyn or Flushing/Navy are prime candidates.

  • It’s impossible to overstate what a disaster those “mixing zones” are, for multiple reasons. Drivers essentially use them like freeway slip roads, as places to lose a bit of speed while still taking the turn at high speed. I note someone suggests that all cyclists veer round cars blocking the mixing zone but, as someone else comments, that’s not always possible. Lines of cars form, blocking the way through, and there doesn’t seem to be a single driver in the whole damn city who understands the meaning of the “yield to cyclists” signs. The problem’s further exacerbated by many drivers’ habit of using the start of the mixing zone as a convenient spot to park/make a delivery/send that text message. The stopped vehicles obscure drivers’ view of riders moving up the bike lane. They then veer into the sidestreet at the last minute, cutting across the unfortunate cyclists.

    It feels far safer, of course, to go up streets with dedicated cyclist phases, like 8th or 9th avenues. The problem there, of course, is that the cyclist phase is short and not timed to allow cyclists a convenient “green wave”. The arrangement is essentially an invitation to cyclists to run most of the lights or get nowhere.

    My preferred arrangement would be one with, perhaps, a continually flashing yellow for drivers turning across the bike lane, bike lanes with physical separation up to as close to the junction as possible and lanes for turning cars that force drivers to turn sharply and pay attention.

    Sadly, however, I doubt that will be done and we’ll be stuck with more deadly mixing zones.

  • Alexander Vucelic

    My Observation is native NYr drivers tend to be more courteous in the miting zones than the dreaded Car with NJ or FLS plates. I also observe Yello Cab drivers looking very hard in the rear view mirrors to check if cyclists are coming.

    We cyclists do need to recognize that the mixing zone design is rather counter intuitivs for drivers with basically zero visability to the protected bike lane. We should also recognize that we are very difficult to see amidt the background visual clutter.

    These observations in no way excuse the killer driver whipping around the corner at highway on ramp speeds.

  • This one’s a great big d’uh. All you have to do is take a newish rider out and see their faces at the first mixing zone.

  • Bobberooni

    The data shown in this article show no benefit of split phase signals for pedestrians or drivers, because the data are just not shown. It would be nice to see if the data ACTUALLY show this benefit. That would be a good argument for trying them out, even at risk of safety problems that come by shortening the pedestrian time.

    I also wonder whether better biker training might help here (yea, that elusive biker training, which we currently have NO effective infrastructure to deliver). Riding through mixing lanes require a completely different approach from split phase signals. How many bikers are even aware of the difference? How many bikers know how to look over their right shoulder and merge to the right of traffic every two blocks?

  • Bobberooni

    I’m wondering… why has the 8th Ave bike lane been so ineffective at reducing risk for bikers? I have some hypotheses, but real data would be nice.

  • Bobberooni

    It’s already the law in NYC that you don’t have to use the infrastructure if you don’t want to. The problem is hot-under-the-coller auto drivers who don’t understand that, who think that bike lanes are for the convenience of drivers, and who feel that bikers not in bike lanes are fair game for harassment.

  • Bobberooni

    I agree. Riding to the left of a car in a left-turn lane is incredibly unnerving. I always merge to the right of the left-turning traffic, spilt-phase or no. I’d never heard this was illegal, but certainly have never been stopped by a cop for it.

  • Simon Phearson

    I don’t think so. See Section 4-12(p)(1) of the NYC traffic rules (NYC Rules, Title 34, Chapter 4), which states, “[w]henever a usable path or lane for bicycles has been provided, bicycle riders shall use such path or lane only,” except when necessary to leave the lane to turn or to avoid obstructions that make it unsafe to use the path/lane.

  • Moe

    This sentence above summarizes what’s in DOT’s report, also linked above: “On average, crashes with injuries declined 30 percent on the six “split-phase” redesigns and 13 percent on the six “mixing zone” redesigns.” The report shows that most of the injury crash reduction is for pedestrians and drivers. Absolute cyclist injury reduction is largely flat because numbers of cyclists increase dramatically on these streets in most cases.

  • Moe

    The 8th Ave lane replaced an existing buffered bicycle lane (which had already done part of the risk-reduction job) whereas cases like 9th Ave or 1st Ave were installed on big highway-like streets that had seen no previous design modification at all

  • walks bikes drives

    I’d estimate that, of all the assholish driver behavior I experience as a cyclists, 8 of ten times it is a driver with out of state plates. 6 of ten are NJ, and the other two are either CT or PA. Anecdotally, the worst are NJ BMW’s, followed by NJ Mercedes.

    Edit: I should add I grew up in NJ and learned to drive there. It was drilled into us twenty years ago that you had to drive defensively no matter what, unless you are driving in NYC.

  • Alexander Vucelic

    We should Be somewhat tolerant Of drivers from NJ and similar exurban landscapes. It must Be utterly bewildering for them to Drive in a real city.

  • Of course, mixing zones would be a very bad choice for the situation where you have a two-way cycle track.

  • Bobberooni

    Not true. See:


    Summary: you can ride outside the bike lane whenever a pedestrian or car is blocking it ==> in Manhattan, you can ride outside the bike lane whenever you like.

  • Bobberooni

    Thanks. But this brings up a disturbing possible conclusion. Does it mean that protected bike lanes are only marginally safer for cyclists than buffered? If so, then StreetsBlog and TA would need better justification to advocate for protected lanes.

  • Simon Phearson

    Well, you’re free to misinterpret the law in whatever way you deem convenient, but the cited provisions clearly do not amount to broad permission to ride outside the lanes whenever you like, not even in Manhattan.

    As I noted, above, the permission is tied to what’s “unsafe” in the bike lane. That might well contemplate avoiding the odd pedestrian or two (in my experience they are not as frequently encountered in the lanes I use as they might be, say, on Eight in midtown), but it certainly wouldn’t include riding outside the lane freely, based on some vague expectation that pedestrians are eventually likely to be using the bike lane. I would expect the provision to be a bit like the RCNY’s exception allowing drivers to drive in the bike lane, which they are free to do when there is insufficient room in the main traffic lane to proceed. They aren’t permitted to drive in the bike lane freely, just out of the expectation that one of those ubiquitous double- or parallel-parkers or delivery trucks might stop in front of them. They’re free to do so when and if those actual obstructions present themselves.

  • BBnet3000

    Safety isn’t the only justification, its just the only justification being used in NYC right now.

    Others include: increased comfort for existing users and more likely to encourage more users (aka subjective safety) and less frequent blockage by double parked cars.

  • BBnet3000

    That arrangement sounds a lot better, and is much closer to how it is done in the most bike friendly places. I suspect that continuing the green pavement across the intersection would help a lot as well.

    The best way to do this in the long term would be by using pigmented asphalt that will not wear off the road. I have no idea whether the additional cost of this while already doing a repaving has ever been explored, and I’ve never seen it mentioned on Streetsblog.

  • J

    True, but they can do this because in Amsterdam the minor street intersections are designed with raised crosswalks, so turning onto them is more like turning onto a driveway: drivers must mount the sidewalk and bike lanes to cross them and reach the side street.

    In nyc, we treat every intersection like a major intersection and we can’t fathom reducing automobile connectivity. As such we’ve created major intersections every 200 feet and in saying that they either be designed well as major intersections (protected intersection) or designed as minor intersections (driveway treatment). Mixing zones are not good designs for any intersection.

  • SFnative74

    Unless it means improving 20 intersections in a period of time with an improvement rather than only 2.

  • chekpeds

    That was me last year….

  • chekpeds

    Well said . We just need to speak loudly whenever DOT proposes bike lanes

  • Joshua Putnam

    You wouldn’t need the awkward term “split-phase signals” if someone hadn’t popularized misuse of the term “protected” when applied to bicycles.

    A “protected” left turn lane isn’t a left turn lane surrounded by walls, it’s a left turn lane that gets a separate signal phase, protected from conflicting movements. A “protected” pedestrian crossing is a signal that doesn’t allow any conflicting vehicle movements while pedestrians are allowed in a crosswalk.

    But a “protected bike lane” as the term is popularly misused *is* a bike lane surrounded by barriers, and it doesn’t necessarily have any protection from conflicting movements.

    Perhaps what this article really means is that there’s a meaningful distinction between *protected* bicycle lanes and *separated* bicycle lanes.

  • Safety benefits may be marginal, but the ridership gains are far more substantial.

  • Joe R.

    Right, and I’ve argued for exactly the same thing in NYC, namely to treat minor side streets more like driveways. Bollard them off at one end so they can’t be used for through motor traffic. This way only the people who live there or make deliveries there will need to go down the street. You’re 100% right. We don’t need the level of automobile connectivity we currently have.

    Mixing zones are not good designs for any intersection.

    As Robert Wright already stated, they’re a disaster. Driveway treatments of minor intersections and protected major intersections are the way to go. Or even better have a bike overpass or underpass at major intersections to avoid conflicts altogether (and also to avoid the need to stop at red lights).

  • Well-designed protected intersections really don’t have much of a deviation from a straight line at all and they’d work quite well with 30′ cross streets.

  • Joe R.

    I think the problem isn’t with protected bike lanes themselves, but rather that there are specific situations where they make sense, and others where the benefits are marginal. The best place to put protected bike lanes is on a street which has no motor cross traffic on one side for long stretches (i.e. PPW). Cyclists can just yield to peds while motor traffic is stopped at red lights. Since there is no turning or cross motor traffic, you have long stretches where the cyclist has virtually no conflicts, other than needing to watch for pedestrians in crosswalks.

    When you put protected lanes on streets with intersections every 250 feet, then arguably they fail to provide much protection at the one place you need it the most, namely intersections. And the mixing zones only add to the confusion and danger. This isn’t to say there isn’t still some benefit, like preventing parked cars from encroaching on the bike lane, but the benefits are minimal compared to using a protected bike lane in a place where one really makes sense.

  • Ziggy Tomcich

    While I’m a big fan of the protected intersection, one of the key elements of them mentioned in the video is a dedicated signal phase for cyclists. Still, this is a much more sensible and obvious solution to intersection design than the bone-headed mixing zone.

  • AndreL

    Countries with heavy urban cycling such as Netherlands and Denmark have no problems with cycle-specific signal phases. There is absolutely nothing wrong having specific phases for cycling. We need to move past this dismissal of traffic segregation of bicycles and cars.

  • jsallen

    Not mentioned in the discussion: split signal phases for bicyclists and motorists reduce the green time, increasing delay and increasing the temptation for bicyclists and motorists — not only pedestrians — to disobey traffic signals. The problem is less serious where cross streets are one-way, as on 9th Avenue in Manhattan, so only 1/2 of the streets have turns across the bikeway — but could be handled better if bicyclists were encouraged to merge into the slow or stopped left-turning motor traffic — and then would get as much green time as motorists. Compare 9th Avenue with the very poor performance of the mixing zones on Grand Street, also in Manhattan. More details on both of these installations are at http://john-s-allen.com/galleries/NYC/

  • Joshua Putnam

    Interesting reading from Denmark, where a “mixing zone” model is one of the alternatives recommended for reducing right-hook fatalities.

    Note that they recommend starting the mixing zone much earlier than most U.S. installations, 15-25m (50-80 feet) before the stop line, and that drivers should have 70 m (231 feet) of sight distance *back* down the cycletrack *before* they begin merging to the right.

    The rearward visibility is important to allow drivers to see bicycles approaching the mixing zone before driving into it, something that’s very difficult when parking is continued nearly to the merge point, as in the illustration for this story.

    It would be interesting to see results in the U.S. for mixing zones providing a more adequate merging distance and greatly increased rearward sight-lines for drivers.


  • Joshua Putnam

    I’ve been surprised at the number of people on bikes willing to run the red bicycle lights on Seattle’s 2nd Ave cycletrack. (Not anywhere close to a majority, but a larger minority than I would have expected.)

    Technically, that’s legal in Seattle since the city hasn’t yet recognized bicycle signal faces in the municipal code and the Legislature hasn’t recognized them at the state level, but I would have expected most people to respect the red bicycle light out of simple self-preservation, even though the wait can be interminable if you hit multiple blocks of red lights.

  • jsallen

    PPW is Prospect Park west in Brooklyn.

  • jsallen

    There is no such thing as subjective safety. Safety is measured in crash rates, not in what people think they might be. Call it “the appearance of safety” and please remember that appearances can be deceptive.

  • Joe R.
  • Joe R.

    The problem here isn’t the concept of split phase signals. I know they exist overseas. Rather, it’s the frequency which we’ll use them. In general traffic signals on bike routes are avoided as much as practical in places like the Netherlands. I think the tendency is to consider one signal per km roughly as a maximum guideline, except in rare cases where you can time a sequence of closely spaced signals to match average cycling speeds. In the US we take their concept of split-phase signals, but like virtually everything else we do regarding traffic controls, we overuse it. A split phase signal every other block for the mixing zones, plus one every block for the cross streets, along with no chance of timing anything to match cycling speeds, ensures cyclists will pretty much ignore red lights. Once a significant number of users ignore a traffic control, it ceases to have any effectiveness. They already realized long ago in the Netherlands that cycists don’t like to stop. Therefore, they designed their bike routes to keep stopping to a minimum. They’ve even built elaborate underpasses or overpasses to get cyclists past busy intersections without stopping. If there are traffic lights, more often than not they exist to give bikes priority over motor traffic. The rare times cyclists might need to stop occur because there was likely no cost effective way to engineer out the conflict without split phases.

  • Alex Brideau III

    A good point, though strictly in lay terms I’d argue that the earlier use of the term “protected” is the more misleading than the latter, as a lane “surrounded by walls” or other barricades can actually provide physical protection, whereas an exclusive phase of a traffic light cycle provides no actual physical protection, just legal protection.