Cycling impact
A bike lane moves people on a sliver of the space a car lane eats. Below is Toronto's measured cycling network and ridership, next to a bounded scenario of what ripping those lanes out would actually do to car traffic. Spoiler: removal usually makes congestion worse, not better.
- Cyclists/day, counted
- 35,633trips a day across the counted network
That is real, measured demand, not a projection. It rides on 839 km of cycling network made of 1,581 mapped segments, clocked by 43 permanent bike counters.
- Network length
- 839 km
- Cycling segments
- 1,581
- Bike counters
- 43
If these lanes were removed scenario
Removing lanes displaces the 35,633 counted cyclists/day. Some would drive instead. The share who switch to a car is uncertain, so we show a bounded range sourced to published cycling mode-shift studies, not a single number. Each row is the added car travel that switch would create (their own trip time; it does not count the extra congestion they add to other drivers).
| Mode shift to car | Added car trips/day | Added person-hours/yr | Added cost/yr |
|---|---|---|---|
| 15% | 5,345 | 650,302 | $12,030,592 |
| 30% | 10,690 | 1,300,605 | $24,061,183 |
| 50% | 17,817 | 2,167,674 | $40,101,972 |
The lane cars would get back
scenarioThere is a real argument for removal, so here it is with the numbers behind it. Take out a separated bike lane and its curb space goes back to general traffic, so drivers get an extra lane. An extra lane genuinely eases congestion, and this section estimates how much rather than pretend that gain is zero. The table is that gain: the driver delay an added lane would save each year across the 110 km of on-street cycle track, and what the saved time is worth. Off-road trails are left out because they do not sit beside a road lane, so removing them frees nothing for cars.
Why a scenario and not a measurement? Our routing engine assumes traffic always flows freely, so it never sees congestion and cannot measure congestion relief. Instead we compute the relief with the standard traffic-engineering method (the Highway Capacity Manual and the Bureau of Public Roads delay formula) under stated assumptions: 10,000 vehicles a day in each direction, 35% of them in the busy peak, and a road that widens from 2 lanes to 3 when the bike lane comes out. Change those assumptions and the gain changes, which is why we show a range.
| Peak congestion (volume/capacity) | Driver delay saved (person-hrs/yr) | Value of that time/yr |
|---|---|---|
| low (0.75) | ~140,000 | ~$2,600,000 |
| mid (0.90) | ~290,000 | ~$5,400,000 |
| high (1.05) | ~540,000 | ~$9,900,000 |
Each row is a busier road. Volume/capacity is how close it runs to full at peak: 0.75 is busy, 0.90 is near capacity, and 1.05 is over capacity (stop and go). The more congested the road, the more an added lane helps, so the saved time grows down the table.
So does removal come out ahead? No. Even in the mid case, the lane handed back to cars saves about 290,000 person-hours a year, but the new driving from displaced cyclists (30% of them switching) adds far more: roughly 1,000,000 more person-hours a year even after crediting that saved time. The two sides are not a tidy ledger, since one counts former cyclists' own added driving time and the other counts relief to the drivers already there, so read the net as a rough illustration. Even granting the strongest version of the case for removal, removal still adds more delay than it saves.
Safety
measuredAcross Toronto, 0 collisions that killed or seriously injured someone (KSI) involved a cyclist. This is an observational comparison: it describes where injuries happen, not a controlled causal estimate.
| Facility | Cyclist KSI | Length | KSI / km |
|---|---|---|---|
| Separated (cycle track / trail) | 0 | 490 km | 0 |
| Non-separated (lane / sharrow) | 0 | 350 km | 0 |
Read the per-km rates with care: much of the “separated” length is off-road multi-use trail, which sees few motor-vehicle collisions by its nature, so the near-equal rates are partly a denominator effect, not evidence that separation doesn't help. The share of collisions happening away from any bike facility, below, is the more informative signal.
Of all cyclist-KSI collisions: 0% occurred on a separated facility, 0% on a painted/shared facility, and 0% away from any bike facility.
Exposure-normalized rate unavailable: insufficient counter coverage near corridors.
What we don't model (yet)
Reclaimed-lane capacity is now modelled above as a scenario. The gap that remains is induced demand. When you free up road space, some of it fills back in with new driving trips that would not have happened otherwise, and those trips eat into the reclaimed benefit. So treat the net figure as an upper bound on the real relief.