Today we’re publishing our first annual report: The State of Urban Cycling Infrastructure 2026: A global benchmark of 750 urban areas.

It asks three plain questions about every one of those 750 urban areas, from places of 52,000 residents to cities of 43 million:

  1. Provision: how much cycling infrastructure is there, and of what kind?
  2. Mapped continuity: does it join up into a connected network?
  3. Proximity: how many people live near the lowest-stress routes?

It’s written first for city planners, and it’s free to read, reuse and check.

What we found

Here are the headline figures. Each one comes with the areas it describes, because a number without its denominator is easy to misread.

  • 379 of the 750 areas have too little low-stress cycle network mapped to judge whether it joins up. That’s a rule about evidence, not a finding that those places have no infrastructure.
  • In the median area, there are 0.07 km of mapped cycle infrastructure per km² of built-up area (all 750 areas). The highest is 8.95.
  • 42% of all mapped cycle kilometres are cycleways whose form the map doesn’t record, so how much of the network is separated from traffic is still unresolved.
  • Where continuity could be measured (366 of the 750 areas), the median network keeps only 24% of its low-stress length in its largest connected piece. Not one of those 366 networks is a single connected piece.
  • Counting every resident of the 750 areas once, 16% live within 300 m, in a straight line, of a lowest-stress cycle route. In the median area it’s 4%.

Two more patterns stood out:

  • Outer neighbourhoods are usually served a little less well than central ones, but in the median area the gap is under one percentage point.
  • Smaller areas are withheld far more often: 21% of areas with 50,000 to 99,999 residents had enough mapped network to compare, against 69% of areas with a million or more.

Near is not the same as connected

One idea runs through the whole report: living near a cycle route isn’t the same as being on a network. A short protected lane that connects to nothing still counts as “near” for the homes around it. That’s why the report measures proximity and continuity separately, and why closing a short gap between two pieces of network can matter as much as building new kilometres.

How we made it

  • Open data, every figure traceable. The measures come from OpenStreetMap and global population grids (GHSL and WorldPop), extracted between July and November 2026.
  • 750 areas chosen on purpose. They’re a deliberately selected benchmark, not a representative sample: almost one in three has fewer than 250,000 residents, and more than a third are in Asia.
  • Insufficient evidence isn’t a zero. An area where the map is too thin is reported as insufficient evidence rather than scored.
  • Checked, and honest about how. The findings passed automated checks and two reviews carried out by an AI model, not a person. The edition hasn’t been externally reviewed yet, and the report says so.
  • The evidence is public. Every area’s outcome and the evidence behind it are in a downloadable evidence package, so anyone can recalculate each figure without asking us.

Read it, reuse it, cite it

The report’s text, data tables and graphics are under CC BY 4.0: share and adapt them for any purpose if you credit Party Onbici and link to the edition. Map-derived data stays under the Open Database License, © OpenStreetMap contributors.

To cite it:

Joseph, A. (2026). The State of Urban Cycling Infrastructure 2026: A global benchmark of 750 urban areas. First edition. Party Onbici. doi:10.5281/zenodo.24270250

The report and its data are archived at Zenodo under that DOI.

Tell us what we got wrong

This is a first edition, and we’d rather hear about a mistake now than repeat it next year. If you know a city in the report, check its result and send us a correction, a question about the method or what you’d need from the next edition. There’s a feedback form at the end of the report, and you can comment on any section.

Read the 2026 report, and if it’s useful, pass it on to someone who plans streets.