Starting January 29, 2020, Market Street became closed to cars from 10th to Main in the eastbound direction, and from Steuart to Van Ness in the westbound direction. This closure was part of the "Better Market Street" project, designed to improve Market Street for people walking, biking, and taking public transit. Without private vehicles on Market Street, buses would be less impeded, and people walking and biking would be at lower risk of getting into collisions with cars. This analysis focuses on the latter goal, by investigating if closing market street to cars improved safety strictly for cyclists.
This analysis draws on multiple data sources to assess changes in bike safety and ridership on Market Street.
Traffic Crash Data
Comprehensive data on traffic collisions was obtained from the SF Gov “Traffic Crashes Resulting in Injury” dataset. This dataset includes police-reported crashes citywide involving injuries to motorists, pedestrians, or cyclists. It provides details such as crash type, location, date/time, and parties involved.
Bike Volume Data (SFMTA Counters)
Ridership trends were analyzed using the SFMTA’s automated bike counter data. These counters track the number of bicycles passing fixed locations throughout the city, including Market Street.
Street Use Permits
A list of issued and approved street use permits was obtained from the SF Department of Public Works "Street Use Permits" dataset. This dataset includes any permit submitted to use space on the sidewalks or streets, from setting up tables and banners, to large excavation projects. It provides details on location, and duration the permit is approved for, as well as what the permit is for and who it is issued to.
The traffic crash dataset only includes crashes reported by SFPD. There are likely more bicycle related crashes that went unreported so it should be assumed that this analysis overestimates actual bicycle safety. The dataset includes traffic crashes reported between January 1st, 2005 to May 31st, 2025. Bicycle crash data was extracted from the full crash dataset by identifying records where a bicyclist was listed among the involved party types. Only crashes occuring on Market Street, and within the boundaries that are closed to cars, are selected.
The street permit data set is geo-filtered to only include permits issued within the span of car-free Market Street. Also, since construction permits that may encroach on the road and/or bike lane are of interest, only the following permit types are included: ["TempOccup","Excavation","ExcStreet","StreetSpace","Emergency","StrtImprov","Wireless","AddlStSpac","SideSewer"]
Bicycle crash data from Market Street is shown below from 2015 through 2024. This gives a five year sample of crashes on the same section of Market prior to closing the street to cars, and a five year sample during the closure. The timing of the closure roughly coincides with the beginning of the pandemic, in which bicycle traffic patterns changed drastically. SFMTA bicycle counter data from the 9th-10th cross street counters is used to estimate the number of bicycle rides occuring on Market Street, and is shown in another figure below. This data has several key limitations such as missing rides beginning or ending east of the counters, and not counting bicyclists who pass by a counter outside of the lane. Additionally, bike counter data only begins in 2018, giving a smaller sample size for pre-closure analysis.
Following the closure to cars in January 2020, the number of Bicyle-involved crashes declines dramatically--comparing the five year period before vs. during shows a 55% decline in crashes. Furthermore, a montly high of five crashes is observed only once with Market Street closed to cars, while in the five years leading up to the closure there were nine months with five or more crashes. However, bicycle traffic also reduced heavily. Comparing a two year period before the closure to the five years that follow, shows a 41% decrease in the average yearly bicycle count. This is most likely due to the pandemic disrupting in-office work and associated commuting patterns. Yet the frequency of crashes per ride is reduced following the closure of Market Street to cars. Crashes-per-10,000 counted rides is shown below, and comparing the two measured time periods reveals a 35% reduction in annual average crash frequency.
| Period | Year | Number of Bike Crashes | Average per Year |
|---|---|---|---|
| Pre Car-Free (2015-2019) |
2015 | 28 | 35.0 |
| 2016 | 26 | ||
| 2017 | 37 | ||
| 2018 | 51 | ||
| 2019 | 33 | ||
| Total | 175 | ||
| Car-Free Period (2020-2024) |
2020 | 11 | 15.6 |
| 2021 | 13 | ||
| 2022 | 24 | ||
| 2023 | 12 | ||
| 2024 | 18 | ||
| Total | 78 |
| Period | Year | Total Bike Rides | Average per Year |
|---|---|---|---|
| Pre Car-Free (2018-2019) |
2018 | 1,330,804 | 1,337,074 |
| 2019 | 1,343,344 | ||
| Car-Free Period (2020-2024) |
2020 | 842,214 | 784,702 |
| 2021 | 721,301 | ||
| 2022 | 840,858 | ||
| 2023 | 729,794 | ||
| 2024 | 789,342 |
| Period | Year | Crashes per 10,000 Rides | Average per Year |
|---|---|---|---|
| Pre Car-Free (2018-2019) |
2018 | 0.38 | 0.31 |
| 2019 | 0.25 | ||
| Car-Free Period (2020-2024) |
2020 | 0.13 | 0.20 |
| 2021 | 0.18 | ||
| 2022 | 0.29 | ||
| 2023 | 0.16 | ||
| 2024 | 0.23 |
After Market Street was closed to cars, there were substantial shifts in both the types of crashes that occur and where they occur. This analysis focuses on two major changes. First, it examines how the locations and causes of crashes between bikes and cars has shifted. Second, it analyzes the change in the most prevalent type of bicycle crash—from collisions involving cars to crashes involving just the bicyclist and no other road users.
When Market Street was open to cars, crashes between cars and bicycles were nearly evenly split between intersections (defined as within 20 feet of a cross street) and the mid-block. At intersections, crashes involving cars proceeding straight were the most common, while mid-block crashes primarily involved cars that were slowing, stopped in traffic, parked, or parking/re-entering traffic. After Market Street was closed to cars, this distribution shifted sharply toward intersections. This change is expected: with far fewer vehicles traveling along the mid-block (the street remains open to taxis, buses, and government vehicles), opportunities for mid-block collisions are significantly reduced but still possible, while cross streets continue to introduce risk. Prior to the closure, 50.6% of car–bike crashes occurred at intersections; this figure has since increased to 81.2%. The types of car actions preceding a crash at intersections has remained largely consistent with the pre-closure period. Most crashes still involve a car proceeding straight, and the share involving a turning vehicle has remained nearly unchanged—23% during the car-free period compared with 22% before. Given that cars are significantly less likely to be turning onto and off Market Street, the persistence of this pattern is notable and may warrant further investigation into how these crashes are classified or reported.
When Market Street was open to cars, crashes involving a car and a bicyclist made up the largest share of bicycle-related crashes. After the street was closed to cars, the share of these crashes declined from 46.3% of all bicycle related incidents to 36.8%. Bicyclist-only crashes—-those involving no other road users-—became the most common incident, accounting for 49.4% of all bicycle-related crashes.
Despite becoming the most common type of collision, the frequency of bike-alone crashes was lower during car-free Market Street. 2020, 2021 and 2022 had the lowest rates of bike-alone crashes per 10,000 rides, while 2024 matched the rate observed the year prior to closing Market Street; however, in 2022 this rate skyrocketed, and was significantly greater than the highest observed crash rate before Market Street was closed. This bump remains apparent when looking at all bike related crashes per year, but particularly in 2022, the percentage of crashes that were bicyclist-alone jumped to 62.5%, as compared to 46.1% in 2021 and 41.7% in 2023. Further investigation of 2022 is presented later on, and analyzes another factor impacting bicycle safety.
Bicycle-alone crashes on Market Street have a significantly higher percentage of severe injuries reported than crashes between bikes and cars--19.1% of bicyclist-only crashes resulted in severe injuries, compared with 3.5% of those between bikes and cars. This difference seems counterintuitive, but it could be attributed to a reporting bias. Incidents in the data represent crashes that Police have responded to, so it's very likely Police will respond to incidents involving multiple road users, than a single road user, unless the latter was a more severe crash.
Since the majority of crashes during car-free Market shifted to bicycle-alone crashes, the overall crash severity outcomes during car-free Market Street will appear to be more severe. However, comparing the severity of bicycle-alone crashes in the five years prior to five years during car-free market street, shows a decline in the percentage of severe outcomes (21.6% to 16.1%).
Bicycle Alone Crashes Pre Car-Free Market Street: Total=51, Severe=11, Percentage=21.57% Bicycle Alone Crashes Following Car-Free Market Street: Total=43, Severe=7, Percentage=16.28%
In 2022 there was a noticeable rise in bicycle crash frequency, increasing from 0.18 crashes-per-10,000 rides in 2021 to 0.29 crashes-per-10,000 rides in 2022, and then back down to 0.16 in 2023. As mentioned earlier, a significant majority of these crashes were bicyclists-alone. The types of crashes observed in 2022 and 2023 are compared below, and its immediately apparent that the concentration of mid-block crashes was significantly higher in 2022. Additionally, the "overturned" category of crash the second leading type of crash in 2022--with the majority of them occuring in the mid-block, but this category is completely absent in 2023.
One hypothesis for an increase in collisions, especially those involving bicyclists-alone, is that the road obstructions and construction projects became more prevalent and created a more challenging and dangerous path for bicyclists. To assess this hypothesis, street use permits deemed likely to have encroached on car-free Market Street were obtained, and active permits per month and per year are plotted below. Remarkably, there is a significant increase in active permits during 2022! The number of yearly active street use permits and total bike crashes are strongly correlated with a p-value of 0.009. On a monthly timebase, this correlation is less statistically significant, with a p-value of 0.07, but still non-negligible. Further analysis on this hypothesis is warranted, and could include spatial correlations along with temporal.
Correlation Analysis: Pearson correlation of monthly DPW active permits to Crashes per 10,000 rides: 0.20, p-value: 0.070 Pearson correlation of yearly DPW active permits to Crashes per 10,000 rides: 0.88, p-value: 0.009
My name is Carlos, I frequently bike on Market street to commute. I’m here to share some concerns I have regarding ride sharing services coming to market street —specifically how it will negatively impact bicycle safety. The proposal includes an evaluation period where you’ll be looking at safety metrics, but during this period ride sharing will only be operating at limited destinations and at off-peak hours. This means you’re not going to get a representative assessment of the safety implications. Also, it’s unclear what the rubric for safety assessment is, like how many people need to be injured for this proposal to be reconsidered? This feels like more of a formality than a legitimate attempt at understanding the safety implications of allowing ride sharing on market street. Instead, if you could look back to when cars were allowed, and you’ll see some pretty revealing metrics. So prior to closing market street to cars 53.4% of crashes between cars and bikes occurred in the mid block area so not at an intersection but ON market street - 134 crashes from the time span available. Since closing market, that’s been reduced to 19% or just 6 crashes, with the rest happening at intersections. But we should expect this. But what’s really striking about this is that of those prior mid-block crashes, 44% of them involved a vehicle that was either stopped in the road, parking, parked or re-entering traffic — All things that Ubers and Lyfts do very frequently and very unexpectedly. So reintroducing ride sharing services on market will re-enable this crash mechanism as well as countless other mid-block crashes that have nearly been eliminated. Putting some numbers to this, even if the rate of mid block crashes from before is halved you’re still looking at a 30% chance 1 or more cyclists will be injured in a crash with a car per month. So the mayor has provided no evidence that granting waymo, uber and Lyft permission to operate on market street will revitalize downtown. Instead it seems he would rather cut muni service and endanger residents who choose to walk and bike. This is not transit first and is not what this board should be approving. Thank you
road_cond_1 No Unusual Condition 34 Other 12 Not Stated 3 Obstruction on Roadway 2 Name: count, dtype: int64 intersection Intersection <= 20ft 30 Midblock > 20ft 21 Name: count, dtype: int64 party1_dir_of_travel East 32 West 16 Not Stated 2 South 1 Name: count, dtype: int64
/var/folders/ry/6zlsyrrj22zffwv1bszfkwz80000gn/T/ipykernel_9437/2543725821.py:12: DeprecationWarning: *scattermapbox* is deprecated! Use *scattermap* instead. Learn more at: https://plotly.com/python/mapbox-to-maplibre/
/var/folders/ry/6zlsyrrj22zffwv1bszfkwz80000gn/T/ipykernel_9437/2860319746.py:15: DeprecationWarning: *scattermapbox* is deprecated! Use *scattermap* instead. Learn more at: https://plotly.com/python/mapbox-to-maplibre/ /var/folders/ry/6zlsyrrj22zffwv1bszfkwz80000gn/T/ipykernel_9437/2860319746.py:37: DeprecationWarning: *scattermapbox* is deprecated! Use *scattermap* instead. Learn more at: https://plotly.com/python/mapbox-to-maplibre/
before car-free after car-free Top 10 crash locations before and after car-free Market Street: Before Car-Free Market Street: address MARKET ST 05TH ST 16 MARKET ST 07TH ST 15 MARKET ST 04TH ST 13 MARKET ST 08TH ST 12 MARKET ST 02ND ST 10 MARKET ST 03RD ST 9 MARKET ST 06TH ST 8 MARKET ST CYRIL MAGNIN ST 7 MARKET ST 11TH ST 7 MARKET ST SUTTER ST 7 Name: count, dtype: int64 After Car-Free Market Street: address MARKET ST 05TH ST 11 MARKET ST 07TH ST 10 MARKET ST 04TH ST 7 MARKET ST MCALLISTER ST 5 MARKET ST MASON ST 4 MARKET ST 01ST ST 3 MARKET ST 08TH ST 3 MARKET ST 02ND ST 3 MARKET ST SUTTER ST 3 MARKET ST STOCKTON ST 3 Name: count, dtype: int64