Riding a Bike is Pretty Safe
The data on cycling risks in Palo Alto, California and Pasco County, Florida
“Mike,” a friend said, “I liked your post on cycling safety,” looking away slightly. Then he looked me in the eye. “But it made me wonder if I should stop riding my bike.”
I wrote recently that bike fatalities increased by 50% in a decade — a large and unfortunate increase. And the infrastructure for biking in the US leaves much to be desired. But in the grand scheme of things, riding a bike is pretty safe.
In this post, I’ll outline a framework for thinking about health and safety risks. Then I’ll explain the relative risk levels of a few activities, and how I came to the conclusion that riding a bike is a pretty safe thing to do.
Defining Risk Levels
How can we quantify the risk of sudden fatality from things like car crashes? My approach is to look at three numbers:
Yearly Deaths Per 100,000 Population: How many out of 100,000 people (for context, Mountain View, Redwood City, and Palo Alto all have a little less than 100,000 residents) will die from a given cause in a year? e.g., it could be the case that 4 out of 100,000 people will die in a certain kind of accident.
Lifetime Deaths Per 100,000 Population: How many out of 100,000 people will die this way over their lifetimes? Note that this is someone’s entire life, not just what they have in front of them and implicitly includes people who recently were killed via this type of accident and likely would otherwise still be alive.
Yearly Deaths Per 100,000 Participants: This is a measure of how many participants are killed doing a particular activity. For an activity like being in a car or being a pedestrian — where nearly everyone takes part — this number will be quite similar to the population level. For a non-universal activity like riding a bike, it will be higher.
I won’t calculate lifetime participant deaths per 100,000, because participation in activities tends to change over time. I didn’t start riding a bike until I was 13 (I know, I know!), and I probably won’t be riding on the road when I’m 80 even if I’m still active and fit (I’ll explain why).
American Road Risks
Let’s start with the overall risk from car crashes.
Yearly Deaths Per 100,000 Population: In a typical American city of 100,000 people, about 13 people will be killed in car crashes each year. Just over 2 people (about 2.2) will be pedestrians. And once every four years, a bike rider will be killed (a rate of about 0.25).
Lifetime Deaths Per 100,000 Population: In a typical American city of 100,000 people, about 1000 (1%) will be killed in a car crash over their lifetimes. About 175 (0.175%) will be pedestrians killed by car drivers. About 20 will be bike riders killed in accidents (0.02%).
Yearly Deaths Per 100,000 Participants: Most Americans do not ride a bike in a given year, and it’s surprisingly difficult to get an accurate number on how many people ride. Statista claims 54.7 million while a Breakaway Research Survey shows over 100 million. If we use a slightly more conservative figure of 50 million, we can calculate that about 1.8 out of 100,000 bike riders each year are killed in the US. I won’t calculate this number separately for people who ride in cars and pedestrians, as virtually everyone in the US is in each of these categories.
I’ll add a point of comparison to bike riders: motorcycle riders. There are about 8.6 million motorcycles registered in the US, and in 2021 there were about 4000 people killed in motorcycle crashes. Assuming that the number of motorcycles and motorcycle riders are about even, that would mean that there are about 45 motorcyclist fatalities for every 100,000 motorbike riders — a rate 25 times as high as the fatality rate for cyclists.
Road Risk in the Area Around Palo Alto
I live in Palo Alto, a weird (mostly in good ways) bubble of technology and nice weather and wealth and better than average bike infrastructure. That leads to risk that’s pretty different from the national averages.
To understand my risk, I looked at the area around Palo Alto, roughly between San Carlos and Cupertino, which has 23 ZIP codes1 and 675,000 people.
This area had 370 people killed in car crashes in the 15 years between 2007 and 2021.
The metrics for this area are fairly different from national averages:
Yearly Deaths Per 100,000 Population: The typical American city of 100,000 people will see 13 people killed in car crashes each year; in my region, only 3.7 per 100,0002 are killed. Just one person per 100,000 in my region is killed as a pedestrian (less than half the national average). The rate of bike riders killed here is about the same as the national average — around 0.25 per 100,000 residents.
Lifetime Deaths Per 100K Population: While in the typical American city of 100,000 people, about 1000 (1%) will be killed in a car crash over their lifetimes, in my area it will be about 300 out of 100,000 or 0.3%. About 80 people (0.08%) will be pedestrians killed by car drivers. About 20 bike riders will be killed in accidents.
Yearly Deaths Per 100,000 Participants: There is better data on cycling in my area: in a recent survey, 12.9% of people reported riding a bike in a typical week in Santa Clara County3 (disclosure: a company I co-founded, Change Research, conducted that survey). Riding a bike in the last week is a significantly more aggressive definition of a bike rider, but I’ll use it here. 12.9% of people riding a bike translates to 88,000 riders in this area, or a yearly death rate of about 2 per 100,000 active riders. That means that for an average active (weekly) cyclist in Palo Alto, the risk from being in a bike crash is about the same as the risk of being a driver or passenger in a car crash. Both numbers are based on averages, and don’t account for the timing and safety of any individual’s driving or biking.
Road Risk in Pasco County, Florida
Pasco County, Florida is a county north of Tampa, with a population slightly smaller than the area I looked at. Like Palo Alto, it has good weather for most of the year, which tends to encourage bicycling. It’s a bit more rural and significantly more red than Silicon Valley is. And its roads are a lot more dangerous.
Yearly Deaths Per 100,000 Population: My area has an overall car crash fatality rate of about 3.7 per 100,000 people. Pasco County’s fatality rate is about 18 per 100,000 people — about five times where I live. The pedestrian fatality rate is 3.5 per 100,000 people — more than four times Silicon Valley’s rate, and about the same as the total car crash fatality rate for my area. There are about 1.5 bike rider deaths per 100,000 people in Pasco county — over five times the rate in Silicon Valley.
Lifetime Deaths Per 100,000 Population: Pasco’s higher yearly crash fatality rates add up. Overall, about 1300 people (1.3%) will die in car crashes. About 250 people (0.25%) pedestrians will be hit and killed. And over 100 bike riders will be hit and killed.
Yearly Deaths Per 100,000 Participants: Unfortunately, I don’t have statistics on bike riding in Pasco County, so I’m not comfortable providing an exact number here. However, given the higher rates of bicyclist fatalities in Pasco County, this number is likely appreciably higher than it is in and around Palo Alto.
Where and When Bike Crashes Happen in Palo Alto and Pasco County
Risk is not uniform, and that applies to riding a bike in both Palo Alto and Pasco County.
In the area around Palo Alto, 28 cyclists have been killed in the past 15 years. 22 have been male, and six female; that’s actually more evenly split than the rest of the country (87% of bike riders killed are men). Fatalities are fairly spread out in terms of day of week (7 of 28 on weekends) and time of day (19 during the day, 6 and night, 3 at dusk or dawn).
The largest bloc of bike riders killed around Palo Alto have been older: the median age of cyclists killed in this area is 59.5 and the most common ages of victims are bike riders in their 60s and 70s.
Half of cyclist deaths around Palo Alto were on roads classified as local streets; a few each were on county roads and state highways, and just one was on a US highway.
Pasco County tells a very different story. Like Silicon Valley, the vast majority of bike riders killed are male, and there’s no major difference by day of the week. But almost everything else is different.
The median age of bike riders killed in Pasco County was 48.5, and the most common age range is 40s and 50s. Though Pasco County has had far more bike fatalities than Palo Alto since 2017, there have been more cyclists over 60 killed in Palo Alto than in Pasco County.
In Pasco County, slightly more fatal crashes happened at night than during the day. Around Palo Alto, three times as many crashes happened during the day.
Perhaps most strikingly, the vast majority of fatal crashes in Pasco County happened on a county road, a state highway, or a US highway. Local streets — the most common location of fatalities around Palo Alto — comprised less than 10% of fatal bike crashes in Pasco County. It’s likely that the road types where deaths occur in Pasco County have cars going at higher speeds.
Data I Wish I Had
There’s a lot of data that — to the best of my knowledge — isn’t available. It would be amazing to have:
Fine-grained data on all bicycle crashes, not just fatalities
Detailed data on the cyclist: what kind of bike were they riding (mountain bike, road bike, e-bike, etc.), were they wearing a helmet, did they have lights, were they riding on the right side of the road?
Good data on cycling behavior: how many people are riding in a given location on a given day or at a given time of day?
More data on transportation behavior would be immensely valuable for assessing risk accurately. There are too many cases where we have the numerator (e.g., the number of bike riders killed) but not the denominator (e.g., the number of miles ridden on bikes). This is true for many other types of road deaths as well.
My Personal Approach to Risk
Here’s my summarized approach to risk, as a middle-aged man who lives in Palo Alto and rides a bike semi-regularly is:
The risk to me of riding a bike is roughly in line with that of driving a car in Palo Alto, and it’s significantly lower than the risk of driving a car in many parts of the country. If I ride for ~50 years of my life, my overall risk of dying in a bike crash is about 1 in 1000 — higher than I wish it was, but something I’m willing to accept in exchange for the freedom and fitness benefits of riding a bicycle.
I can mitigate this risk by riding as safely as possible: preferably during the day and with good lights at night, riding predictably on the right side of the road, and wearing a helmet.
I was struck by the higher accident rates for older riders in Silicon Valley. That will certainly make me think twice about continuing to ride into my 60s. At a minimum, I’ll spend more time looking at the data as I get older and my reaction times slow down.
I feel fortunate that I can mostly ride on local roads around Palo Alto, which seem to be safer. There is certainly a ton of room for improvement with bike infrastructure.
I’m not sure that I would feel the same way if I lived in a place like Pasco County; even if I were to ride a bike, I would be reluctant to ride on highways and at night.
I would also be more wary of driving in Pasco County given the relative risks.
I do not ride a motorcycle — the risks are too high.
We Need Better Cycling Infrastructure
Like many Americans, I’ve had the experience of visiting European cities like Copenhagen and seeing bike infrastructure that is miles (or kilometers) beyond what we have in the US. This can be great from both a lifestyle perspective and a safety perspective: Denmark has much higher rates of bicycling than the US (16% of all trips are done on bike) and yet there are only around 20 cycling fatalities per year across the whole country.
I’m immensely grateful for the American spirit of technology innovation — this area in particular has built amazing companies and amazing technology. When it comes to innovating and upgrading transportation infrastructure — notably for bicycles — the US has much room for improvement.
For my analysis, I looked at 94301, 94303, 94305, 94306, 94043, 94041, 94040, 94042, 94027, 94061, 94025, 94022, 94028, 94304, 94024, 94087, 95014, 94085, 94086, 94089, 95051, 94063, and 94070.
The area where I live is roughly as safe as Germany and Spain, it’s safer than Canada, Australia, New Zealand, Italy, France, and South Korea, and it’s less safe than the UK, Japan, and Scandinavia. Considering how much wealthier my area is than those countries — the area around Palo Alto has multiple trillion dollar companies and none of those countries have any — it still feels like we are underachieving on safety.
Santa Clara County is a slightly different geography than the one I looked at, but there is substantial overlap and I believe that the numbers would be similar for the region I am examining.