Cyclist Compliance at Red Lights in Minneapolis

Watching a cyclist fly through a red light while we wait idling in our car, on our feet or on our own bikes is frustrating. Such a brash move in the face of society causes us to lambast the enforcement of the legal system, we criticize their parents’ child rearing skills, and start attributing recklessness to the entire race of cyclists. But do all cyclists disobey stoplights? No, of course not! On my daily rides I often observe cyclists stopping and waiting for the green, and I know several cyclists that habitually obey all traffic laws. So, how many scofflaws are there and what attributes and behaviors do they share? Nearly two years ago, I set out to find these answers and in the hope of adding facts to the often heated & opinionated discussion of cyclists and their propensity to stop at red lights (this post has been a long time coming). Below is a distilled version of my much lengthier study.

My initial search for similar published studies unearthed two which looked at the compliance of cyclists at red lights. However, neither study took place in the U.S. (Australia and China instead) and both involved reviewing video footage of intersections in order to identify specific behaviors and attributes of cyclists. Because I didn’t have the resources to invest in video cameras, nor the time to review footage, Jacob Thebault-Spieker (a UMN PhD student in the Department of Computer Science and Engineering) helped me customize a mobile app he developed called CitizenSence to meet my data collection needs. With the app I was able to quickly and easily record real-time observations of individual cyclists in the field and beam the data to a central server which automatically formatted it into a downloadable Excel file for analysis. Easy peasy!

Deciding to record cyclist observations with a phone app rather than video sped up the collection and data analysis process and it enabled identical parameters for observations to be replicated across different intersections (and users if desired). But the app’s interface and my own limitations (sadly I only have one set of eyes and I do not have x-ray vision to see cyclists through big trucks) did narrow the amount of data I could collect. So, of the 18 variables the other two other studies collectively addressed, I determined eight could feasibly be recorded with the existing CitizenSence interface. Then those eight variables were dissected and combined to create the final attributes below:

  1. GENDER – Both studies found women to have higher compliance rates than men
  2. STOP TYPE – Each cyclist observed was categorized as having stopped then proceeded through the red, waited for red to turn to green before proceeding or failed to stop
  3. DIRECTION OF TRAVEL – Direction of travel was entered as proceeding left, straight or right through an intersection
  4. GROUP SIZE – This measure attempts to account for behaviors related to group size by comparing the number of members of the stopped group and the stop types for each person within that group
  5. MISCELLANEOUS: GPS coordinates and the time at which each observation was made were recorded automatically with each submission (each submission being an individual cyclist) using the mobile app. Also other intersection characteristics such as types of bike facilities and number of traffic lanes at each observed intersection were recorded


I chose 15 signalized intersections across Minneapolis with high cyclist volumes recorded by the City of Minneapolis during recent signal retiming projects. Six of the 15 chosen intersections happened to also be listed within the top 33 intersections with the most bike / vehicle crashes from 2000 to 2010 as stated in the Minneapolis Crash Report. Over the course of my study, I counted each intersection twice resulting in 30 separate counts.

I visited the intersections and made observations whenever I was available on a weekday between 3pm and 7pm (evening rush hour was chosen in an attempt to capture the highest daily volume of cyclists). In all, 9.5 hours of observations over eight fair weather days (no rain and above 60F) were recorded between May 6th and June 28th, 2013.

Of the 411 cyclists I observed, 239 (58%) complied, meaning they came to a complete stop and waited for the light to turn green before proceeding. Interestingly, the proportion of non-compliant cyclists ranged from 8.7% to 68.9% across the 15 different intersections.


My overall findings showed that:

  • Nearly half (47.7%) of the observed cyclists actually encountered red lights
  • The majority of both genders (70.3% of females and 52% of males = 58.2% overall) complied
  • When turning right, the majority of both males and females failed to comply with red lights (88% and 81.3% respectively)
  • Of the 349 cyclists traveling straight 63.3% complied while 57.1% of the 21 cyclists traveling left complied
  • Those riding alone were compliant 46.2% of the time compared to groups of two (76.8%), and three (80.9%)
  • Cyclists at intersections with sharrows or bike boulevards had the worst and second worst compliance rates (31.1% and 37.2% respectively)
  • Gender varied most at intersections with sharrows (males: 89%, females: 11%) and was most similar at intersections with separated bike paths (males: 55%, females: 45%)


I also found that compliance levels varied dramatically by intersection:

  • Minnehaha Parkway & Portland Ave had the highest compliance rate (91.3%) of any intersection observed
  • Lasalle Ave & W 15th St, and Lake St & Bryant Ave had the worst compliance (31.1% and 37.2% respectively) and had the greatest number of cyclists that stopped but proceeded through the intersection (24 and 20 respectively)
  • W 24th St & Nicollet Ave, and Franklin Ave & Portland Ave had the greatest number of cyclists that didn’t stop at all (18 and 17 respectively)
  • Lasalle Ave & W 15th St had the greatest rate of red light cyclists per minute (1.56) compared to an overall average of 0.64 per minute

Overall, Minneapolis cyclists comply with red lights 58% of the time, which is right in the middle of the Australian study (7%) and the Chinese study (79%) compliance rates, but well below the Portland study (which didn’t include right turning cyclists in their 94% compliance percentage). While it’s comforting to know that most cyclists here do obey red lights, the importance of this study and future ones is to open a dialogue about what influences the compliance of all road users. Is it a function of intersection complexity (number of lanes, speed of traffic, types of bicycle facilities) and how confident people are to take risk yet be safe? Or is compliance a function of the presence or absence of a clear and attractive option to comply (designated spaces, nice way-finding all the way through the intersection)? Or is it something else entirely?

Ultimately, compliance is good, but safety is better. Don’t get me wrong, law enforcement is important, but ticketing red light runners doesn’t provide better cycling conditions (unless the generated fees went directly towards building better cycling facilities). Instead, engineering and educational pursuits like Minneapolis’ Safety Starts With All Of Us, have the potential to yield better results. By informing and empowering the public, narrowing travel lanes, reducing speeds, and building designated facilities for each mode that minimize the number of conflict points (where paths cross), people will know where their vehicle belongs and trust who has the right of way. They will know that the coil buried beneath the road will detect them and trigger the light to turn green. They will know how to act predictably. After all, we should strive to be a society where someone riding a $30 used bicycle is provided the same level of safety and security as someone driving a $30,000 car. And using data from monitoring cyclists behaviors and attributes to identify problems, promote solutions and safeguard present and future citizens is a fantastic step in that direction.

About Michael Petesch

My work is at the nexus of environmental and transportation planning. I’m strongly interested in creating community scale wellness through the planning, deployment, management and evaluation of multimodal transportation, multifunctional public spaces and the integration of nature throughout urban landscapes. I bike. I camp. I eat.