Health and Evacuation Challenges for Transit Riders During the 2025 Los Angeles Wildfires

Publication Date: 2026

Abstract

Click to View the Research Brief

Wildfires pose threats to health due to exposure to wildfire smoke and other toxic debris. Those who rely on public transit are potentially more exposed to these threats, yet little is known about how transit riders navigate this increased exposure amid evacuation and displacement. In this study we examine the exposures of transit riders during a fast-moving wildfire emergency, the 2025 Los Angeles wildfires. We conducted a mixed methods study of transit riders, drawing from a survey and semi-structured interviews of riders sampled from Transit, a widely used real-time bus tracking app for transit riders. Preliminary findings show that during the wildfire, many transit riders relied on informal rides, public transit, or walking to evacuate, leading to longer evacuation times and heightened exposure to smoke. Survey-linked air quality data revealed that nearly half of Black and Latino respondents experienced “unhealthy” PM2.5 concentrations at their homes on peak smoke days, a burden echoed in widespread reports of respiratory and mental health impacts. Riders adopted protective strategies such as wearing masks and avoiding transit when possible, but those without cars faced limited options. These results underscore the need for transportation planners and emergency managers to design evacuation and communication strategies that explicitly support transit-reliant and car-less households.

Introduction

In January 2025, a rapid sequence of wildfire events devastated dense, urban areas across Los Angeles (LA) County. Between January 7 and January 31, 2025, amid an extreme Santa Ana wind event, the Palisades Fire burned roughly 23,448 acres, while the Eaton Fire scorched about 14,021 acres and destroyed or damaged more than 18,000 structures. Over 200,000 residents were placed under evacuation orders, with more than 100,000 receiving direct evacuation mandates. In addition to the structural damage, these urban wildfires caused extreme smoke and air pollution resulting in major chaos in the lives of people across Los Angeles County.

Fast-moving wildfires pose significant challenges to evacuation, especially for transportation-insecure households who are more likely to rely on public transit, walking, or bicycling to meet their daily needs. Disruptions in public transit not only hinder the ability of transit riders to evacuate, but they also limit access to essential services. As transit services were canceled, rerouted, or delayed, transit riders faced unreliable mobility amid extended exposure to hazardous particulate matter (PM2.5) pollution. Despite the severe risks present and the convergence of wildfire, toxic air pollution, and urban transit disruption, research on how transit-reliant populations perceive, respond to, and adapt in such emergencies remains virtually nonexistent. This study addresses this critical gap.

Leveraging a rider survey collected via the Transit app and interviews with transit-riding evacuees, we focused on transit riders impacted by January 2025 wildfires including the Palisades, Eaton, and Sunset fires. For the purposes of this study, transit riders included users of any of LA Metro’s transit modes: bus, bus rapid transit, or rail. Our research pursued three interrelated objectives:

1. Document how transit riders evacuated during the January 2025 wildfires.
2. Evaluate transit riders’ exposure to toxic air pollution while evacuating and in the emergency and temporary sheltering periods.
3. Investigate how transit riders adopted health‑protective behaviors (wearing masks or avoiding travel and transit, etc.).

By highlighting the experiences of Los Angeles’s transit-riding populations amid unprecedented wildfire smoke and mobility disruptions, this study offers needed evidence to inform equitable, health‑conscious emergency and evacuation planning for wildfire-prone urban areas. The growing frequency and intensity of wildfires necessitates evacuation systems that account for urban populations without reliable car access.

Literature Review

Transportation Insecurity and Disaster Vulnerability

Researchers have defined transportation insecurity as the inability to travel between places in a timely manner due to insufficient transportation resources (Murphy et al., 2022¹). This condition has been particularly acute for older adults, people with disabilities, and low-income households, who are more likely to rely on public transit, walking, or bicycling for their daily needs (Palm et al., 2024²). In emergencies, these populations face compounded threats: they cannot easily switch to safer or faster modes of travel and often encounter extended exposure to hazards. For example, during the COVID-19 pandemic, wealthier transit riders shifted to driving or walking, while riders with fewer mobility resources continued to use transit, risking exposure to reach essential destinations such as their jobs and/or to receive medical care (He et al., 2022³; Palm et al., 2021⁴). In the context of hurricanes, transportation vulnerability has historically been defined based on access to a vehicle, with research demonstrating that carless individuals experience major challenges to evacuation (Renne et al., 2009⁵). Despite the similar risks that transportation insecure people face during wildfires, wildfire evacuation and transportation insecurity have remained largely underexamined in disaster research.

Wildfire Evacuation Behavior and Decision-Making

Research on wildfire evacuation behavior provides policymakers with insight into factors influencing evacuee actions, including risk perception (Forrister et al., 2024⁶; Grajdura & Rowangould, 2025⁷), notification of evacuation orders or warnings (residing within an evacuation zone, receiving a mandatory evacuation order, etc.) (Folk et al., 2019⁸; McLennan et al., 2019⁹), transportation options (driving, walking, getting a ride, etc.) (Kuligowski, 2021¹⁰; Zehra & Wong, 2024¹¹), and departure timing (the time at which an evacuee begins evacuating) (Grajdura et al., 2021¹²; Wong et al., 2023¹³). However, most studies focus on vehicle-owning households, often assuming that evacuees would use their private vehicles. A smaller body of research has examined outcomes for households with reduced or no vehicle access, showing that these households experienced greater evacuation delays, constrained destination options, and increased exposure to hazards (Grajdura et al., 2022¹⁴).

Transit-based evacuation research has remained limited, though case studies have offered valuable lessons. Wambura and Wong (2024¹⁵) found that transit riders in Alberta, Canada needed physical and medical support, multilingual communication, and fare suspensions. During the 2021 Caldor Fire, the Tahoe Transportation District evacuated more than 1,800 people on buses by utilizing a disaster registry, fare waivers, relaxed baggage limits, and partnerships with community organizations (Goddard, 2024¹⁶). While informative, these examples are rare. Despite the lack of empirical data, researchers have designed simulation tools and algorithms to model how agency could deploy buses for evacuations during emergencies (Feng et al., 2023¹⁷; Shahparvari et al., 2019¹⁸).

Transit Riders’ Exposure to Air Pollution During Wildfires

In general, research finds that bus riders experience higher exposure to particulate matter than car or subway travelers (Cepeda et al., 2017¹⁹; Van Ryswyk et al., 2021). In the context of wildfires, transit riders’ health may further suffer, as wildfire smoke has been linked to increased mortality and cardiovascular morbidity (Alahmad et al., 2023²⁰; Chen et al., 2021²¹). There is some evidence that wildfire smoke events can trigger shifts in the types of transportation people take. For instance, Han and colleagues (2024) found that Washington D.C. Metrorail ridership increased by 3.4% during smoke episodes, likely due to pedestrians seeking perceived protection. However, this research used aggregated ridership data, which limits understanding of individual-level behavioral changes. To our knowledge, no studies have directly measured wildfire smoke exposure among transit riders during evacuations.

Research Questions

The evacuation choices of transit riders during fast-moving wildfires are not well understood. While some research suggests that transit riders are more exposed to air pollution than car drivers, the ramifications of this exposure during wildfire events is not understood, both in the context of evacuations and in relation to wider regional air quality impacts. To address these knowledge gaps, this study answered the following questions:

1. How did transit riders evacuate from the Palisades and Eaton fires?
2. To what extent were transit riders exposed to toxic air pollution while evacuating and during the immediate aftermath?
3. How did transit riders perceive air pollution exposure and engage in health-protective behaviors? What factors shaped their decisions?

Research Design

The study used a mixed-methods approach that combined surveys and interviews to understand the evacuation behavior of transit riders and explore air quality impacts. We chose a mixed-methods approach to gather different data types to produce the most comprehensive insights possible. Our research team first deployed a survey through the transit app to reach as many Los Angeles County transit riders as possible. We then asked respondents if they would be willing to participate in follow-up interviews to share their experiences. Our final sample included 623 completed surveys (collectively among evacuees and non-evacuees) and 35 interviews.

Study Site and Context

Los Angeles County

Los Angeles County is a densely populated urban area in Southern California with a population of 10 million people spanning a 4,000 square mile area. Most of the population resides in the Los Angeles Basin, which is flanked by vast wildlands such as the Santa Monica Mountains National Recreation Area (245 square miles) and the Angeles National Forest (1,094 square miles), which comprise one-third of LA County alone, in addition to the Hollywood Hills, Griffith Park, and other wilderness areas at various levels of development. The combination of an urban metropolis and expansive wilderness areas creates a large population living within the wildfire-urban interface (WUI). An increasing amount of Los Angeles County is being classified as very high fire hazard severity zones, according to the California State Fire Marshall (Los Angeles Fire Department, 2025²²).

Los Angeles 2025 Wildfires

Before the fires began, weather forecasters predicted a historically strong Santa Ana wind event to hit the Los Angeles Metropolitan region. Record-high winds reaching 50 to 80 miles per hour, combined with low humidity and dry vegetation conditions, created extremely high risks for the ignition or spread of wildfires in the region.
On January 7 at approximately 10:30 a.m., the Palisades Fire sparked in the Pacific Palisades area in a portion of the City of Los Angeles in the foothills of the Santa Monica Mountains. By 12:00 p.m., the first mandatory evacuation orders were issued for residential areas. At approximately 7:00 p.m., much further east, the Eaton Fire sparked in the Altadena community, an unincorporated portion of Los Angeles County in the foothills of the San Gabriel Mountains north of the City of Pasadena. Mandatory evacuation orders were first issued at 7:30 p.m., with many areas receiving orders overnight. By noon on January 8, both fires had rapidly spread, with each fire reaching roughly 11,000 acres in size and zero percent contained. At 6 p.m. that evening, the Sunset Fire sparked in the Hollywood Hills, and mandatory evacuations were issued for the area. In contrast to the Eaton and Palisades fires, the Sunset Fire only reached 60 acres, and the evacuation orders were lifted the following day as the fire was contained. Nearly 200,000 residents were under evacuation orders throughout the region (Siegler, 2025²³). Ultimately, these fires were among the most catastrophic in the history of California, leading to nearly 17,000 structural losses and 30 reported fatalities (CalFire, 2025²⁴). However, modeling of excess deaths during January 2025 predicts that an additional 440 people may have died from direct effects like lung or medical conditions exacerbated by the fires or indirect effects like lack of medical care or mental health impacts (Paglino et al., 2025²⁵).

Sampling Strategy and Data Collection

The study used a mixed-methods approach combining surveys and interviews. The survey component focused on collecting data on people’s self-reported health effects, travel behavior adaptations, and evacuation process. The Qualtrics survey was distributed through the Transit app which is a widely used transit trip-planning application. LA Metro also promotes the app on its website and uses it to survey its own ridership. For our survey, we placed a bilingual (English and Spanish) banner ad in the app from February 3 to February 12, 2025. The banner was shown to anyone using the app in Los Angeles County. As shown in Figure 1, the app also sent a push notification on February 11 to people who had actively engaged with the app but had not clicked the banner. The banner reached a total of 233,214 app users, with 21,807 people (9.4%) clicking on the banner. The push notifications were sent to 82,655 people, with 2,326 people (2.2%) tapping on the notification. To ensure appropriate participation, an eligibility criterion was applied so that only individuals aged 18 years or older could complete the survey. In total, 1,386 people started the survey, of which 623 provided valid responses.

Figure 1. Transit App Banner Ad and Push Notification

At the end of the survey, respondents were asked if they were willing to participate in further data collection. Those who agreed were subsequently contacted for follow-up interviews to gain deeper insight into the decision-making process surrounding when, where, and how to evacuate, as well as how their lives had been affected since the wildfire. All interview participants were evacuees, and 35 interviews were conducted via Zoom.

Survey Measures

The survey comprised four main sections: housing, health, travel behavior, and demographics. Across these sections, two cross-cutting themes guided survey question design. The first was a comparison between the pre-wildfire evacuation period and the post-wildfire evacuation period, enabling analysis of changes in conditions, health, and behaviors. The second was an emphasis on how transit usage was affected by wildfire, with particular attention to the intersection between transit use, health-protective behaviors, and health impacts. In addition, the survey content was adapted slightly based on respondents’ evacuation status. Those who reported evacuating were asked additional questions about their evacuation process details such as travel mode used, evacuation destination, and transportation difficulties associated with evacuation. This branching structure ensured that both evacuee and non-evacuee experiences were captured, while maintaining comparability across the core survey domains.

Interview Guide

For the interview component, protocols for each participant were tailored following a close review of their survey responses. This allowed the interviews to examine individual experiences more deeply as well as clarify survey-reported information. While a structured interview guide was used to ensure consistency across interviews, the approach was highly flexible, enabling interviewers to adapt questions and follow emerging lines of discussion based on the context of each participant.

Survey and Interview Sample

Table 1 presents the socio-demographic profiles of the survey respondents and interview participants, respectively. To assess the representativeness of our survey sample, we compared respondent demographics against transit rider profiles reported in LA Metro’s own onboarding survey (Los Angeles Metro, n.d.²⁶). The two datasets present a ridership profile that is plurality Latino and majority very low-income. However, our sample is whiter and has a smaller share of very low-income households compared to Metro’s. This can be attributed to the areas affected by fires, including Altadena, West Hollywood, and Pacific Palisades, having lower shares of Latino residents and higher incomes compared to LA Metro’s service area as a whole.

Table 1. Demographics of Survey Participants and Interviewees

Data Analysis Procedures

To analyze the survey data, we first cleaned the data, then conducted exploratory data analysis including visualization, crosstabulations, and descriptive statistics in R Statistical Analysis software. Interviews were recorded and transcribed using Zoom. The research team employed a structural coding approach (Saldaña, 2021²⁷), developing an initial coding scheme aligned with the domains of the interview guide. We conducted codebook calibration to refine the code definitions, followed by intercoder reliability trial to ensure agreement among all coders. Each interview was coded in Atlas.ti by two authors and ties were broken by a third coder. Given the scope of this report, we employed a condensed version of thematic analysis, focusing on identifying the most prominent themes. While this abridged analysis ensures alignment with the report’s aims, a more comprehensive thematic examination will be conducted and presented in forthcoming papers.

Ethical Considerations and Researcher Positionality

The human subjects research protocol for the survey was approved for exemption by the University of California Los Angeles (UCLA) Human Research Protection Program while the interview protocol was approved by the Institutional Review Board at California Polytechnic State University. Survey participants received $10 gift cards while interviewees received $50 incentive cash. Data protection procedures included separating identifiers from transcripts and survey responses. Leaders of the research team completed the CONVERGE training on Broader Ethical Considerations for Hazards and Disaster Researchers (Adams et al., 2021), and all investigators had requisite training for human ethics research.

The research team originally planned to engage community organizations in co-developing the survey and interview guides. However, given that local partners were still deeply focused on providing direct relief, the team decided it was best not to introduce new burdens. Instead, several team members with training in trauma-informed research and lived experience of wildfire-related loss revised the instruments to ensure sensitivity. Team members with lived experience also led all interviews with interviewees who lost property during the fires, introducing themselves as both researchers and individuals who had experienced loss from a wildfire to build rapport and trust with participants who had undergone similar experiences. The team layered an interdisciplinary perspective onto our lived experiences of fire, weaving together methodologies and frameworks from transit planning, transportation engineering, air quality analysis, and urban geography.

To ensure our interpretations of results stayed grounded in local knowledge, we presented results at a half-day workshop for practitioner peer review. The event was held at the Southern California Association of Governments on June 26, 2025, and included sixteen staff and representatives from city and county departments, transit agencies, and planning bodies across the region. Researchers overviewed preliminary results and led guided discussions with agency staff on meaning-making and identifying gaps in the data. Out of seven agency participants who completed a post-event evaluation, six strongly agreed that attending the event was valuable. Four strongly agreed that they learned something useful for their work, and three somewhat agreed.

Air Quality Data and Estimated Exposures

We used the U. S. Environmental Protection Agency’s (EPA) AirNow data to estimate the air pollution concentrations that survey respondents were exposed to (AirNow, n.d.²⁸). Survey respondents provided their home address, which was matched to air quality data and used as an approximation of their air pollution exposures. This comes with the limitation that exposures likely happened both at and away from their home address. Even more, people who evacuated their homes were less likely to be exposed at their home addresses. Although the use of static exposure estimates is a limitation of this study, the survey did not collect detailed temporal and geographic locations that could be used to estimate dynamic exposures. The interview results also support that evacuations were highly varied between participants, with some returning home very quickly (within one day) and others waiting extended periods. Including air pollution at the home address of evacuees can also serve as a proxy for the intensity of the fire and smoke circumstances these survey respondents experienced (not necessarily their exposures) and is therefore retained in the sample.

The EPA collects AirNow data from air quality monitors, which are reported hourly. This study used daily average PM2.5 concentrations, given they are particularly harmful to human health (Environmental Protection Agency [EPA], 2025²⁹), and followed AirNow interpolation methods to estimate air pollution concentration contours, which estimate an air pollution concentration at each address. Finally, the daily average PM2.5 concentrations were categorized according to the USEPA’s Air Quality Index categorical breakpoints (US EPA, n.d.³⁰), such that concentrations below 9 μgm-3 are good, concentrations between 9.1 and 35.4 μgm-3 are moderate, 35.5 to 55.4 μgm-3 are unhealthy for sensitive groups, and 55.5. to 125.4 μgm-3 are unhealthy.

Results

How Transit Riders Evacuate During Fast-Moving Wildfires

The largest number of respondents evacuated via rides from others (28%, n=46), followed by public transportation (21%, n=34) and walking (19%, n=31). Transit riders were least likely to evacuate using their own car or ride-hailing or car share as shown in Figure 2. Groups most likely to rely on transit or active transportation during evacuation included African Americans, low-income households, no-car households, and Latino respondents—groups more likely to be exposed to unhealthy conditions during evacuation. Conversely, White and higher-income respondents were the least likely to use these modes, tending instead to evacuate by private vehicle.

Figure 2. Evacuation Mode Choice by Subgroup

Interviewees provided further insight into people’s mode of choice. One person described their thought process: “I was thinking, Uber, if there were any, and if not, like, I would have just biked. Okay, I've biked before with masks.” Over a third of interviewees evacuated by catching a ride with friends, but also recognized that was not necessarily a reliable plan for disasters. In the words of one participant:

So, I have friends in the area with cars, but the area in LA is a pretty big area, so I'm talking [about] people who live an hour or two drive away from me, right? Which doesn't necessarily qualify as an evacuation plan in an emergency.

Another participant was catching a ride with neighbors and got caught in the congested traffic where emergency responders had to instruct people to leave their cars and run due to the encroaching fire. They recounted that:

[We were] told by a guy on a megaphone. I don't know [if] it was police or fire department. It was all so chaotic, get out of your car now, and that's what we did … we carried the three kids down the street onto the beach.

One-fifth of interviewees (n=7) reported using ride hailing service (e.g., Uber, Lyft) for at least part of their evacuation. Another interviewee reported that they were planning to catch a ride but ultimately could not. Without that ride, they were afraid that ride hailing companies would be gouging people. They said:

First the plan was we were getting a ride. So we didn't have to think about [how to evacuate]. We were just trying to gather belongings. But once that fell through and we knew time was of the essence … I just thought if, if the Uber and Lyft is still running, I was scared there weren't going to be drivers or they were going to do really crazy surge pricing.

Despite concerns about price gouging, that interviewee and others were able to successfully evacuate using a ride hailing service, and multiple interviews mentioned the discount for evacuees, although some had difficulty figuring out how to use it. One interviewee said they felt that Uber and Lyft cost “at least twice as much” during the fires.

Evacuation Characteristics

The survey asked evacuees how long it took to evacuate to safety, and over half reported it taking over an hour, as shown in Table 2. Groups most likely to take over an hour included non-binary evacuees, Black and Asian evacuees, and evacuees who did not own or have access to a vehicle. Evacuees least likely to spend over an hour to reach safety included White evacuees and vehicle owners. The survey also asked evacuees if they experienced any difficulties evacuating due to transportation issues, and roughly 39% did (Table 2). Evacuees most likely to report a transportation-related difficulty evacuating included evacuees with disabilities, evacuees without cars, and men.

Table 2. Evacuation Time and Difficulties Among Evacuees

The interviews provided an opportunity to learn more about people’s evacuation experiences. Evacuation challenges were a common theme across all of the interviews. One interviewee was explicit about the incorrect perception of people in Los Angeles all having their own vehicles, including in an evacuation, saying: “there's a stereotype that everyone has a car here, but it's a very exclusionary stereotype, because there's so many reasons why people wouldn't drive or can't drive or can't afford a car.”

One of the most compelling stories among interviewees was one person who could not drive, or get a ride, or get official assistance. They shared that they asked friends for a ride, but:

They couldn't [provide a ride]. Everybody was packed up and ready to go themselves … I called 911 and the 911 operator said that they were stretched so thin that nobody could help me get out … The paramedics were [nearby], and I asked them, could I just have a ride down the hill? And they said, no, they're just stretched too thin … I just needed to get out of this vicinity and get safely away from the fire, the progression of the fire, so I started walking.

People who did evacuate by car reported challenges with traffic congestion, smoke from the fires, and general confusion about where to go or what routes to take. One person described driving on a major street that had become congested: “All the feeder streets along the side, people were coming out. So, there was lots of traffic, there was heavy smoke, so it was kind of difficult to see.” Another participant recalled:

It was just total gridlock … there's two big apartment buildings just north of us, and all the cars were trying to get out of the parking garages simultaneously, and like no one could even get out onto the street. So, it took a really long time just to even get into the flow of traffic.

Several participants noted that, by asking them for suggestions for improvement, the interview had helped them think more broadly about what might be useful for people to evacuate without cars, or with other special considerations like more advanced hazard warnings that might result in a longer evacuation time. One participant shared: “Maybe we need this additional zone that's [for] somebody that maybe doesn't have a car, or has, like, special needs or a family and kids, [maybe there should be] ‘warning plus’ or something.”

Air Pollution and Transit Riders

Survey respondent exposures are quantified as the daily average PM2.5 concentration at their pre-evacuation home addresses for evacuees, and home address for non-evacuees. Pre-fire addresses are used to estimate exposure levels during evacuation, but do not reflect exposure when the respondents completed the survey roughly a month after evacuation. Figure 3 represents the daily average PM2.5 concentrations during the week of the wildfires. Both highlight that peak air pollution occurred on January 8, 2025, the day after the fires started. Figure 3 also contextualizes the pollutant concentrations according to the EPA’s Air Quality Index classifications.

Figure 3. Daily Average PM2.5 Air Pollution Concentrations at Respondents’ Home Addresses During the Week of the Wildfires

The mean respondent exposure on January 8 is classified as unhealthy, and although conditions improved in the following days, the air pollution levels were still considered unhealthy for sensitive groups. These exposures are further explored according to the demographic characteristics of survey respondents as shown in Figure 4.

Figure 4. Air Quality at Respondent Addresses on January 8th by Demographic Group

Nearly half of the Black/African American (47%, n=14) and Hispanic/Latino (49%, n=111) respondents had unhealthy levels of PM2.5 at their residences on January 8th compared to 38% (n=31) for Asian respondents, 42% (n=32) of mixed or other race, 33% (n=42) of White respondents, and 42% (n=263) across the sample. Gender, income, and disability groups have comparable home exposure levels, though low-income participants had notably higher proportions exposed to unhealthy air pollution levels (46%, n=184) than their counterparts (36%, n=79).

The survey also asked participants to report their general self-rated health before and after the fires. Figure 5 summarizes the percentage of respondents who reported a negative change in mental or physical health overall and by demographic group.

Figure 5. Participants Reporting Negative Mental and Physical Health Outcomes by Demographic Group

Changes in mental and physical health were also a common theme in follow-up interviews. Nearly half of the survey participants (48%, n = 298) described physical symptoms related to poor air quality or smoke inhalation, although many participants’ symptoms were temporary and abated in the days and weeks after the fire. Echoing the survey results, several interviewees shared ongoing mental health concerns. As one stated, “I'm literally, like, feel, at times, a little paralyzed, and my brain doesn't work like it usually does.”

Health Protective Behaviors and Barriers to Protective Equipment

The survey also asked respondents what actions or strategies they took to protect themselves from smoke exposure, offering a pre-selected list informed by prior literature. Several options pertained to personal protective measures like wearing masks, which were most used. These included wearing masks (used by 62%, n=388) and drinking more water (35%, n=220). As shown in Table 3, the other protective measures covered travel behavior adaptions, such as switching from transit to Uber, or not going outside at all.

Table 3. Travel Behavior Adaptations Used as Health Protective Measures

The most common travel adaptations were to avoid transit (27%, n=170) or to get a ride (21%, n=128). The least common protective measure was adjusting the time of trips (10%, n=64). Vehicle ownership influenced adoption of protection strategies. Twenty-six percent of participants who owned a car still got a ride, compared to 16% of participants without cars, and 33% (n=97) of car owners avoided transit, compared to 23% (n=50) of non-car owners. Only 6% (n=13) of non-car owners adjusted the time of their trips (compared to 13% for car owners). These results highlight that there are likely different decisions and barriers between car owners and non-car owners.

For some, foregoing all but essential travel proved to be the best response. As one interviewee described: “I tried as much as possible to not be outside until it cleared up, as much as possible to not be in. I think we were advised to not be outside as much as possible.” Masks were an important element to allow people to go out into the smoke, though making them work without a car proved challenging: “even walking to the bus stop I wear an N95 mask…even outdoors, with the N95, you could feel it when you breathe.” As one participant illustrated:

I have a biking mask. That's like, I forget the material, but, yeah, I would like double, triple mask. Like I have this special piece. It's like a silicone or something like that, okay, but it helps. I use it to just help keep them the two masks that I put on my face to keep it from flying off as I'm as I'm biking.

Riders who had to commute described adjusting their routes, accepting longer transit travel times to minimize exposure:

I did change it up. There's like I did at some point I tried to just, like, consolidate the amount of transfers. Like, especially on days where I was feeling extra fatigued and I felt like the air was just out of my lungs, I would just not do the transfer.

Discussion

This study provides one of the first systematic examinations of how transit riders in a major U.S. metropolitan region experienced evacuation, air pollution, and mobility disruption during fast-moving wildfires. Our findings directly address the three research questions guiding this work: how transit riders evacuated, the extent of their exposure to wildfire smoke, and how they perceived and responded to these exposures.

Evacuation Among Transit Riders

Contrary to common planning assumptions that residents will evacuate primarily by private automobile, our results reveal that many transit riders relied on rides from friends, neighbors, or ride hailing companies, while a significant share turned to public transit or walking. These choices were not evenly distributed: Black, Latino, and low-income respondents were more likely to depend on non-automobile modes, which left them vulnerable to longer evacuation times and heightened exposure to smoke and other hazards. Interviews further highlighted the precariousness of these arrangements, with participants describing the unreliability of informal rides, the breakdown of official assistance, and in some cases, the necessity of walking out of danger zones when all other options failed.

Exposure to Wildfire Smoke

Survey data linked to AirNow monitoring showed that nearly half of Black and Latino respondents had “unhealthy” PM2.5 concentrations at their residences on January 8, 2025—the peak smoke day. This exposure burden was reflected in interview accounts of physical symptoms such as respiratory irritation, fatigue, and difficulty breathing, as well as ongoing mental health impacts such as anxiety and depression. While prior literature has documented transit riders’ elevated pollution exposure in routine contexts, our findings extend this evidence to wildfire emergencies, showing that inequities in mobility resources intersect with inequities in environmental health risks. These results suggest that addressing wildfire smoke is not only a matter of air quality management but also of transportation equity.

Protective Behaviors and Decision-Making

Transit riders engaged in a range of protective behaviors, with mask wearing emerging as the most common strategy. Yet adoption of protective measures was uneven: those with cars were more likely to avoid transit altogether, while those without cars had fewer options beyond wearing masks or seeking rides. Riders also wore multiple masks at once, adjusted transit routes, or curtailed trips entirely, often without clear guidance from officials. For transit riders, protective behaviors were not just individual choices but were shaped by access to mobility resources, health conditions, and the availability of timely, trusted information.

Conclusion

Policy Implications

Our findings demonstrate the urgent need for wildfire evacuation planning to move beyond car-centric assumptions and explicitly address the challenges faced by transit-dependent residents. While LA Metro and other large agencies may not be positioned to provide mass evacuation service during fast-moving fire events, policymakers and emergency managers can take concrete steps to reduce risks for riders without cars.

First, agencies can educate transit riders about the importance of having a personal evacuation plan, including identifying trusted neighbors, friends, or family members who could provide rides in an emergency. Outreach materials—distributed through transit apps, community organizations, and multilingual signage—should normalize the idea that having a “ride plan” is as essential as an earthquake kit.

Second, our results highlight the critical role of clear, real-time information on transit disruptions, reroutes, and closures. Riders often lacked reliable guidance during evacuations, prolonging their exposure to smoke. Investment in rapid alert systems—via push notifications, text messages, and neighborhood-level emergency broadcasts—can give riders a better chance of finding safe routes.

Third, protective health strategies must be integrated into emergency planning. Mask distribution at transit hubs, advance messaging about when and how to use protective equipment, and coordination with community-based organizations can help reduce respiratory harm when evacuation options are limited.

Fourth, authorities should adopt targeted alert systems that trigger earlier evacuation warnings for transportation-limited individuals. Proactively identifying households without reliable car access—through opt-in systems, partnerships with community organizations, social service providers, and transit agencies—would allow emergency managers to issue tailored alerts and organize early evacuations for these residents.

Beyond transit focused strategies, policymakers can support car-less riders by preventing ride hail price surges during evacuations. Uber and Lyft provided discounts on rides to designated shelters and evacuation sites, yet many evacuees needed rides to the homes of family members or friends. These riders were subject to surge prices and some evacuees were deterred from this option as a result.

Finally, while this study focused on Los Angeles, its implications extend across U.S. cities at the wildfire–urban interface. Equitable disaster planning requires acknowledging that tens of thousands of urban residents will not be evacuating by car, and their health and safety depend on proactive communication, community coordination, and resource distribution.

Limitations

This study has several limitations. First, our exposure estimates are tied to respondents’ home addresses rather than their actual evacuation routes or sheltering locations. As such, they may not reflect the degree of smoke exposure that occurred during prolonged outdoor travel or time spent in congested vehicles. Second, while the survey sample reflects the general demographic profile of LA Metro riders, it may not fully capture the experiences of the most marginalized riders, including those without smartphone access or English proficiency. Future research should incorporate more dynamic exposure measures and broaden outreach to these groups.

Author Acknowledgments. Additional funding for this study was provided by from the University of California Los Angeles Institute for Transportation Studies Wildfires Rapid Response Research Program. We would also like to thank the following participants of a June 2025 workshop on transit riders and wildfires: Foothill Transit, the Los Angeles Department of Transportation, the Los Angeles Metro, the Los Angeles County Department of Public Health, the City of Los Angeles Department of Emergency Management, the City of Pasadena, Waymo, and the offices of Mayor Karen Bass, Supervisor Kathryn Barger, and Councilmember Imelda Padilla. 

References