One-Way or Two-Way? How to Choose a Protected Bike Lane

TL;DR;

• One-way tracks on each side make movements easier to predict and give direct access to both curbs.¹
• A two-way track can avoid a road crossing, connect to a trail or bridge, and use the side with fewer driveways.²
• Counterflow bicycle movement may be unexpected, so intersections, driveways, and signals must reveal and control it.³
• Research often finds higher intersection risk for two-way tracks, but results are confounded by location, design, and exposure.⁴⁵
• Direction is not protection quality: width, separation, crossings, and maintenance can outweigh the layout.

The choice is not “safe” versus “unsafe”

Imagine a two-way city street. The transportation department has enough room for protected cycling, but two plausible layouts:

1. Put a one-way cycle track on each side, with riders moving in the same direction as adjacent motor traffic.
2. Put a two-way cycle track on one side, with riders meeting each other and one direction moving counter to adjacent traffic.

Montreal’s Saint-Denis REV uses the first layout. The second is common beside waterfronts, parks, bridges, railways, and wide arterials—and formed most of Seville’s rapidly built connected network. Neither arrangement is automatically good.

A cycle track’s directionality says which way bicycles move. Its protection quality says whether riders have adequate width, real separation, clear sightlines, safe crossings, and usable maintenance. Those are different questions. Plastic posts do not become concrete because a lane is one-way; a protected signal does not become unsafe because the approaching track is two-way.

That distinction is the central rule: choose direction at the corridor and network scale, then make every conflict work at the site scale. A design that looks space-efficient in a typical cross-section may create extra crossings at every destination—or an expensive new signal problem at every intersection.

The decision table

These are tendencies, not substitutes for counts, turning diagrams, and observation of how people already travel. They synthesize the MassDOT separated-bike-lane planning matrix, current NACTO protected-bike-lane guidance, and the FHWA separated-bike-lane guide.¹⁶²

Why intersections favor predictable direction

On a conventional right-driving street, a driver leaving a side road and turning right primarily looks left for approaching cars. A cyclist coming from the driver’s right on a two-way track is legal and visible in principle, but outside the movement the driver is preparing to join. A driver turning from the main road across the track must likewise check for bicycles from both directions.

That is not a moral failure unique to drivers. It is a workload problem the design creates. The 11th edition of the US Manual on Uniform Traffic Control Devices says same-direction one-way tracks accommodate road-user expectations and may produce fewer conflicts at intersections and driveways; it flags extra challenges for two-way tracks, including pedestrian interactions.³

Dutch crash research supplies the harder warning. At unsignalized priority intersections, Schepers and Voorham found one-way tracks safer than two-way tracks and found lower crash probability where crossings were raised and the track was deflected 2–5 meters from the carriageway.⁷ A later Dutch synthesis concluded that bidirectional paths increase the directions a driver must scan and may also increase head-on bicycle conflicts.⁴

This does not mean a one-way track can be painted straight through a blind driveway and declared safe. It means a two-way track has a more demanding intersection brief:

• Preserve visibility far enough back from every driveway and corner.
• Mark both bicycle directions through the crossing.
• Slow turning vehicles with geometry, not signs alone.
• Use raised or set-back crossings where the context supports them.
• Separate bicycle and turning-vehicle movements in time when volumes and speeds make concurrent operation unreliable.

The details of corner islands, setbacks, and turn conflicts belong in our articles on why protection often ends at intersections and the right-hook crash. The point here is narrower: bidirectionality increases the number of approach directions the intersection must reveal and control.

Signals are part of the cross-section

MassDOT’s signal guidance uses lower turning-vehicle thresholds for considering time-separated bicycle movements when the track is two-way. Its planning guidance also notes that a two-way facility typically requires additional signal equipment, while a one-way pair may sometimes operate with existing phases.⁸¹

“Sometimes” matters. A one-way track with heavy turning traffic can also need a bicycle signal and a protected turn phase. Conversely, a two-way track beside a road with few crossings may need little signalization between its endpoints. What a city cannot responsibly do is select the two-way cross-section for its apparent simplicity and leave the signal work for later.

What the safety studies can—and cannot—settle

The broad literature leans toward one-way tracks at intersections. A 2013 international review concluded that one-way cycle tracks were generally safer there, while emphasizing that effective intersection treatments improved the performance of protected tracks overall.⁹ Dutch guidance and studies reach a similar conditional preference.⁷⁴

But “two-way” is not a crash cause that can be isolated cleanly from the street carrying it.

Montreal’s classic six-corridor study examined two-way tracks on one side of the road. They carried 2.5 times as many riders as nearby reference streets and had an injury relative risk of 0.72 after accounting for bicycle distance. The comparison was two-way protected track versus no bicycle facility, not two-way versus an equally good one-way pair.¹⁰ It shows that bidirectional protection can be safer than the realistic alternative without proving it is the best possible design.

A 2025 Barcelona analysis made the comparison more directly. Using municipal bicycle counts and injury data from 2020–23, it found slightly higher injury rates on one-way than two-way bike lanes—but a small difference the author judged likely attributable to one-way lanes being concentrated in the central Eixample district.⁵ The dataset mixed physically separated, painted, and sidewalk facilities and did not describe separation or width, so this was not a controlled comparison of protected layouts. Location was doing at least some of the statistical work.

This is why our broader article on protected bike-lane crash risk insists on counting riders. Raw injuries, injury rates, conflict observations, and severity answer different questions. Direction should be recorded as one design feature among many—not used as a label that erases intersection geometry, separation, traffic volume, parking, or signal phasing.

The network can reverse the answer

A one-way pair wins the block-level legibility contest but can lose the trip-level safety contest.

Suppose a school, riverside trail, and most homes all sit north of a six-lane arterial. A two-way track on the north side may let nearly everyone reach them without crossing that arterial. A one-way pair would require half the bicycle trips to cross once to enter the correct track and again to reach the destination. If safe crossings are rare or badly located, people will detour, ride the wrong way, or avoid the route.

FHWA therefore identifies concentrated destinations on one side, connections to other two-way bikeways, and one-way motor streets likely to generate wrong-way riding as contexts where a two-way track may be desirable.¹¹ The older European PRESTO guidance makes the tradeoff especially plain: one-way pairs are normally clearer at junctions, but a two-way track can be justified when it eliminates difficult crossings and makes the network more direct.¹²

Ask four network questions before drawing the lane:

1. Where do trips begin and end? Map schools, shops, stations, housing, trails, and accessible entrances by side of street.
2. Where are the crossings? Count not only junctions but the crossings riders must make to enter, leave, and continue beyond the project.
3. Which side has fewer conflicts? A riverbank with no driveways is different from a commercial curb cut every 20 meters.
4. What happens at both ends? A beautiful two-way track that strands counterflow riders at its terminus merely relocates the dangerous maneuver.

The best side may change along a long corridor. That does not justify weaving the track back and forth whenever curb space gets tight. Each side switch is itself a major crossing and should earn its cost through a clear network benefit.

Width: one barrier does not make half a bikeway

A two-way track can consolidate separators, shy space, and some construction work on one side. It cannot fit two useful directions into the width of one narrow lane.

NACTO’s current guidance gives a minimum rideable width of 6.5–7 feet (2–2.1 m) for a one-way protected lane and prefers 8–12.5 feet (2.5–3.8 m). For two-way tracks, it recommends at least 13 feet (3.9 m) to accommodate varied cycles, passing, side-by-side riding, and platoons; its 8-foot absolute minimum is reserved for short constrained segments.⁶

That generous two-way width matters more as e-bikes, adaptive cycles, trailers, and cargo bikes widen the speed and size distribution. Oncoming riders consume lateral clearance. A faster rider cannot pass a cargo bike whenever the opposing stream is occupied, so a track that looks adequate at low volume can become unstable at peaks. Research on Dutch tracks also finds that riders move closer to the verge when meeting oncoming cyclists and recommends 2.5 meters as a baseline for bidirectional paths in the study context—not a target for a high-volume North American main route.¹³

Capacity is therefore not just bicycles per hour. It is whether a child can ride beside an adult, an adapted tricycle can clear the barriers, and a commuter can overtake without forcing someone toward a curb.

Curb space, pedestrians, and transit

A one-way pair touches both curbs. That can require two sets of loading zones, boarding islands, accessible parking treatments, drainage solutions, and passenger crossings. A two-way track leaves the opposite curb conventional, which can be a meaningful operational advantage.

The cost is concentration. Every passenger, delivery worker, and parked-car occupant using the track side must cross a two-direction bicycle stream. Someone stepping from between parked vehicles may look only toward the direction bicycles usually come from on a road. Clear crossing points, sightlines, tactile information, lighting, and unmistakable two-way markings are therefore essential.

NACTO says curbside loading is compatible with protected lanes but recommends placing loading near block ends where possible; it also calls for accessible spaces and boarding treatments to be planned as part of the project.⁶ At bus stops, two-way or counterflow tracks can further complicate detection for blind passengers. Our floating bus-stop analysis covers that accessibility problem in detail.

The practical curb test is not “how many parking spaces remain?” It is whether the plan identifies where passengers, deliveries, waste collection, paratransit, and emergency access will occur—and gives each movement a legible path that does not turn the cycle track into an informal loading lane.

Winter exposes false economies

A city can build an elegant cycle track in July and discover in January that its plow does not fit between the curbs.

NACTO reports that sweepers and snow equipment range from roughly 4 feet to more than 8 feet wide; cities using wider machines commonly need 7–8 feet of clear access.⁶ A two-way track may be operationally efficient because one passable corridor serves both directions. It also creates a single point of failure: a snow pile, stalled vehicle, construction fence, or drainage problem can block everyone.

A one-way pair duplicates route miles for sweeping and plowing, and crews must service both sides consistently. Yet it may offer more escape points and distribute closures. Either layout needs a maintenance plan, equipment inventory, snow-storage locations, replaceable separator details, and a detour policy before final dimensions are approved. See how winter cities plow bicycle lanes for the operational side of that decision.

Construction cost has no universal winner

The tempting arithmetic says one two-way track needs one separator while a pair needs two, so two-way must be cheaper. Linear barriers, paving transitions, and duplicated curb work can indeed favor consolidation.

But the cross-section is not the project budget. A bidirectional design may require wider reconstruction on one side, new bicycle signals, controller work, turn restrictions, accessible pedestrian treatments, and major transitions at both ends. A one-way pair may need twice the drainage and curb-detail work. Utility locations, existing pavement, bus stops, driveways, and whether the design is paint-and-post or full concrete reconstruction can dominate the direction choice.²

Agencies should publish an itemized lifecycle comparison for both credible alternatives:

• track pavement and separation;
• drainage and utility relocation;
• intersection geometry and signals;
• accessible parking, loading, and transit stops;
• maintenance equipment and annual operations;
• renewal and replacement; and
• safe temporary routes during construction.

A cheaper bid that produces an unusable terminus or an unplowable lane is deferred cost, not savings.

A defensible rule of thumb

Start with a one-way protected track on each side when the street has frequent intersections and driveways, destinations on both sides, heavy pedestrian activity, and a surrounding network organized by travel direction. The familiar movement pattern reduces—but never eliminates—the intersection design burden.

Start with a two-way protected track on one side when that side has markedly fewer conflicts, most destinations lie there, the route follows a park or waterfront, it connects directly to a trail or bridge, or it prevents riders from making one or more dangerous arterial crossings. Fund the signals, visibility, width, pedestrian crossings, and endpoint transitions that make counterflow operation legible.

Reject either option when it is merely the version that fits without moving parking, controlling turns, or buying maintenance equipment. The correct question is not “one-way or two-way?” in isolation. It is:

Which complete design creates the fewest severe conflicts over an entire trip—and remains obvious, accessible, and usable every day of the year?

That answer may be a directional pair on Saint-Denis, a bidirectional spine beside Seville’s road network, or different solutions on different streets. Consistency matters, but honest adaptation to the network matters more.

FAQ

Q. Are one-way protected bike lanes always safer than two-way tracks?
A. No; one-way tracks usually simplify intersections, but protection, crossings, traffic, exposure, and network access can outweigh direction alone.⁹⁵

Q. How wide should a two-way protected bike lane be?
A. NACTO recommends at least 13 feet (3.9 m); its 8-foot absolute minimum should be limited to short constrained segments.⁶

Q. When does a two-way cycle track make the most sense?
A. It is strongest beside few-conflict edges or when it directly serves same-side destinations and avoids dangerous road crossings.¹¹

Q. Is a two-way track cheaper to build and maintain?
A. Sometimes, but savings from one separator can be offset by signal, junction, drainage, transition, and specialized maintenance costs.

References