More Bikeshare Stations is a Good Thing, But It’s Important to Be Realistic

NACTO report

A new study from the National Association of City Transportation Officials (NACTO) says people use bikeshare more when a given area has more stations. But the study makes a density recommendation that’s going to be hard to ever meet, and not everyone agrees it’s a good idea in the first place.

NACTO’s report, released April 28th, adds to the growing body of research that says station density is a key factor in a bikeshare system’s success. While that claim isn’t controversial in itself, NACTO’s suggestions regarding station density cause a bit more friction.

NACTO recommends that cities place bikeshare stations no more than 1,000 feet apart — that is, at a density of 28 stations per square mile. This density would put a bikeshare station within a five-minute walk of each resident in a city.

NACTO’s advice, in fact, is that cities should build out their bikeshare systems at a density somewhere between that of New York’s Citi Bike (the densest system in the U.S.) and Paris’ Velib (the densest system in the world).

The majority of US bikeshare systems are a lot more dispersed than that. Even Chicago, which has received good press for its ambitious Divvy expansion, only plans on a density that’s a fraction of NACTO’s recommendation.


Looking at ridership statistics from bikeshare systems across the U.S., NACTO finds, unsurprisingly, that systems are more successful when they have more stations close together. NACTO says that most bikeshare riders are convenience users, and if a system is not convenient, riders will choose another mobility option.

This has been Washington D.C.’s experience. Before Capital Bikeshare, the city experimented with a precursor known as Smart Bike. Run by outdoor advertiser Clear Channel, the system was largely a failure because it had too few stations and bikes.

Today though, Capital Bikeshare is widely seen as a gold standard. Still, the system only has four stations per square mile, and advocates have called for smaller stations, placed more densely.

Bikeshare systems should fit the populations they serve

But 28-stations-per-square-mile dense? That’s a bit radical, and bikeshare expert Paul DeMaio says it should be taken with a grain of salt. “This proposed station density won’t work well in all settings, such as suburban areas, college campuses, or less dense areas.”

NACTO slideshowA big issue with NACTO’s recommendation is that it doesn’t factor in population density. (For comparison’s sake, Paris is twice as densely populated as New York City, and five times more densely populated than D.C.)

DeMaio maintains that mega-dense station placement can actually have negative effects on a system. “Stations at too high of a density could actually have the unintended consequence of stations cannibalizing trips from the others,” he says. If trips per bike per day is the measure of of a bikeshare system’s success, as NACTO maintains, more bikes and stations regardless of population density can lead to bikes being underused and stations being inactive.  

“Station network density should ideally match the neighborhood density,” DeMaio says.

NACTO says greater station density will not only make bikesharing ubiquitous, but also that it will help jurisdictions address the social-equity problems that have beleaguered bikeshare systems. Low-income areas, according to NACTO, are often built out at a lower density than the system as a whole, making bikesharing a less meaningful option for residents of these neighborhoods.

Bikeshare systems could undoubtedly be denser. More convenient, walkable stations, would increase the usefulness of these systems. 28 stations per square mile, though? This may be a worthy goal, but may also be an unrealistic one for most cities.

Cross-posted at Greater Greater Washington

Photos by NACTO, NY DOT, and Elvert Barnes

Share this item

3 Comments or Mentions

2 Comment(s)


I believe there’s some serious flaws with the methodology used to reach their conclusion of ‘density is king’. There are a few other statements in the document which I believe are false but I’ll stick to the issue of density methodology.

On page 2 the graphs on the right are nonsense. How far is a 15-minute ride (x-axis)? 1.6 miles? That’s half the width of Manhattan. This is simply showing that stations in the center of the system are more used than those on the edge. This is naturally the case and is not related to the 1000 foot density.
The fitted exponential curves are nonsense. Station density and rides per day are not exponentially related – at least not to the extent illustrated.
I think there is some serious confusion between correlation and causation. It is not that stations of high density are highly used but that stations were densely packed/placed in areas of expected high demand!

The methodology presented is flawed. Perhaps there is more that is not presented but it seems that the conclusion is standing on rather shaky legs.

It might be time to send some people to survey residents in socially deprived areas, in proximity to a BSS station, and ask why they don’t use the system.



‘Optimal’ bike share system (BSS) design is a very complicated problem, but there is more to it than simply the station density. One also needs to be very aware of the demand profile and recognize that there are strong directional tendencies (ie the demand in the morning is very different from the evening, as well as during ‘off-peak’ periods and weekends). Furthermore, BSS is frequently a *complementary* mode of travel, not a replacement for one mode or another. That is, a BSS trip is frequently combined with other modes (certainly walking, but also with public transit, and possibly even driving, for example helping people get to a parking garage)

There are a few elements which are reasonably well understood:

1) Walking distance between stations is extremely important. It is very common that one’s first station choice may not be available (see point 2 below) and it is crucial that viable alternatives are available.

2) A station which is either full or empty is only partially filling it’s role in the network. We should strive to minimize the amount of time in which stations are in such an extremal state. Stations along the ‘boundaries’ of the network are most at risk of being in these extremal states, as are stations in ‘high demand’ areas.

3) To help maintain service in high-volume areas, it is useful to have numerous stations in close proximity (say within a block or two). Not all of the stations need to be the same size (as was noted in the report)

5) Convenience is not only measured by ‘distance to a single station’ but also by the number of options within a certain distance (ie the number of stations within a certain radius).



This article has been mentioned in 1 other place(s).