Main

In 2024, the US Surgeon General declared gun violence a public health emergency responsible for tens of thousands of fatal and non-fatal injuries annually, with over 48,000 fatalities in 2022 alone1,2. The increasing burden of firearm violence in the United States has substantial psychological and social ramifications for individuals and communities3, with profound mental health implications that extend beyond the physical toll4.

Existing literature suggests that exposure to gun violence can lead to major psychological distress, including anxiety, depression, panic attacks and post-traumatic stress symptoms5,6,7. Survivors of gun violence, whether directly injured or exposed as witnesses, may experience persistent mental health impacts that undermine their quality of life, including suicidal ideation and psychotic experiences8. Vulnerable populations—including women, younger individuals and racial or ethnic minorities—may face even higher risks for negative psychological outcomes following gun violence exposure9,10. Estimates of post-traumatic stress disorder (PTSD) in survivors of mass shootings range from 3% to 91%, depending in part on how the symptoms are defined11. Communities experiencing mass shootings are often described as co-victims due to widespread social disruptions, including school and business closures, fear and confusion12,13.

While research has begun to explore the psychological impact of exposure to gun violence, substantial gaps in our understanding remain. Many studies have focused narrowly on specific forms of violence, such as mass shootings, which represent only a small fraction of all firearm incidents, or have been confined to certain geographic areas, often overlooking the broader, community-wide consequences of gun violence exposure. The mental health effects of gun violence have not yet been comprehensively characterized on a national scale. In addition, the cumulative effects of different forms of gun violence exposure, such as being threatened with a firearm or witnessing a shooting, are less well documented, especially in nationally representative samples. Most existing research treats firearm violence homogeneously, overlooking the differential impacts based on the nature, proximity and context of the violence14. Some forms of violence may be more consequential for mental health, or some individuals may be more susceptible to psychological effects than others.

These knowledge gaps hinder the development of public health strategies tailored to diverse forms of gun violence exposure. This study seeks to fill these gaps by using a sample of 10,000 respondents, designed to be representative of US adults, surveyed in January 2024 on lifetime exposure to mass and non-mass gun violence and the attendant mental health consequences (Table 1). Beyond recent work that differentiates injury and non-injury exposure to mass shootings15, it introduces community exposure to mass shootings and new analyses on non-mass-shooting victimization, including (1) being threatened with a firearm, (2) being shot at but not struck and (3) being injured in a non-mass shooting. The research specifically investigates psychological outcomes (for example, anxiety, depression, panic attacks and post-traumatic stress symptoms) across multiple forms of firearm violence exposure, ranging from the community level to direct injuries, providing both comprehensive and nuanced insights into the short- and long-term mental health consequences of gun violence. In addition, this study investigates the duration of mental health impacts and identifies patterns of vulnerability among different populations. These findings aim to inform targeted public health interventions that address the needs of both individual victims and the broader communities impacted by gun violence.

Table 1 Descriptive statistics for study participants
Full size table

Results

Gun violence victimization is common in the United States. As shown in Table 2, 20.1% of respondents reported the occurrence of a mass shooting in their community (95% confidence interval (CI): 19.2–20.9%), 4.8% were present on the scene of a mass shooting yet uninjured (95% CI: 4.3–5.2%), and 2.2% were injured during a mass shooting (95% CI: 1.9–2.5%), which could include being shot or trampled or incurring other physical injuries at the scene15. Non-mass-shooting victimization followed a similar trend, with 18.4% reporting being threatened with a firearm (95% CI: 17.5–19.2%), 7.3% were shot at but not struck (95% CI: 6.7–7.8%), and 2.4% were injured in an intentional shooting (95% CI: 2.1–2.8%). These numbers are consistent with national polls of gun violence exposure16 and recent studies in this field17 on lifetime exposure.

Table 2 Self-reported mental health impacts of exposure to various forms of mass and non-mass-shooting gun violence
Full size table

Mental health consequences were common among those exposed to gun violence. Any self-reported impacts ranged from 58.6% (shot at, not struck) to 94.4% (mass-shooting injury). Anxiety/fear was the most frequently reported consequence, ranging from 48.3% (shot at, not struck) to 58.0% (mass-shooting injury). Depression was prevalent among respondents injured in mass shootings, with 74.2% reporting symptoms. Overall, victims injured in mass shootings reported higher rates of depression and panic attacks compared with other forms of gun violence.

The duration of mental health impacts varied. Between 20.3% and 26.3% of respondents who experienced mass shootings in their community or through direct exposure reported mental health distress lasting 1 year or longer. By contrast, about two-fifths of respondents (38.2–42.9%) exposed to non-mass-shooting violence reported long-term mental health impacts, a rate significantly higher than that for mass-shooting exposure. Post-traumatic stress symptoms lasting over a year were reported by the majority of those who were threatened with a firearm, shot at but not struck or injured in non-mass shootings.

As shown in Table 3, age was inversely associated with mental health impacts across all forms of gun violence, with odds ratios ranging from 0.60 (shot at, not struck; 95% CI: 0.48–0.76) to 0.32 (mass-shooting injury; 95% CI: 0.13–0.80). Male respondents were less likely than female respondents to report mental health impacts for community exposure, firearm threats and uninjured shooting incidents. Racial/ethnic differences were observed only for community exposure to mass shootings, with self-reported Black and Hispanic respondents more likely than white respondents to identify mental health impacts. Political affiliation also emerged as a significant factor, with self-identified Democrats more likely than Republicans to report mental health impacts from community shootings, firearm threats and gunshot injuries.

Table 3 Multivariable logistic regression and adjusted odds ratios estimating any mental health impacts of exposure to forms of gun violence on the basis of sociodemographic variables
Full size table

As shown in Table 4, age was positively associated with long-term mental health impacts for non-mass-shooting injuries. Male respondents were less likely to report long-term effects for four of the six forms of gun violence. Black respondents were more likely than white respondents to experience long-term impacts following mass-shooting injuries. Socioeconomic status, political affiliation and engagement with public affairs were also associated with long-term mental health impacts in one or more exposure contexts.

Table 4 Multivariable logistic regression and adjusted odds ratios estimating long-term duration of mental health impacts of exposure to forms of gun violence on the basis of sociodemographic variables
Full size table

Discussion

This study provides evidence that exposure to gun violence, whether through injury or presence on the scene, is associated with profound mental health consequences. The majority of respondents reported psychological distress following gun violence exposure, with anxiety, depression and post-traumatic stress symptoms being the most commonly reported outcomes. While these mental health symptoms were self-reported, they were significantly higher than clinical prevalence rates for the general population, where rates of depression, anxiety, PTSD and panic disorder are 8.3%, 19.1%, 3.6% and 2.7%, respectively18,19. Notably, individuals injured in shootings were most affected, but even those not physically harmed—whether present during the incident or residents in communities affected by gun violence—experienced elevated levels of psychological symptoms. These findings align with previous research indicating the pervasive mental health impact of trauma exposure, irrespective of physical injury5,8. However, our data suggest that gun violence exposure may exert a greater mental health toll than other traumatic events, such as media exposure to the 9/11 terrorist attacks or the stress faced by health-care workers during the COVID-19 pandemic20,21.

The data also reveal that gun violence victimization cuts across socioeconomic boundaries, supporting the conclusion that the psychological toll of being shot is severe and widespread and pervades available resources to mitigate distress. Victims reported high rates of both short- and long-term mental health impacts. Those injured in mass shootings exhibited a higher prevalence of distress compared with those in non-mass shootings, while those exposed to non-mass shootings experienced more prolonged symptoms. These results echo previous findings on the long-lasting effects of trauma, particularly in the context of violent injury6,14,22. Additional research is needed to understand the differences in prolonged symptoms among mass-shooting and non-mass-shooting survivors, which may relate to differences in social support, mental health-care access, the more personal nature of non-mass shootings and exposure to ongoing unsafe environments23,24.

The findings underscore the need for targeted mental health interventions, particularly for populations identified as more vulnerable to mental health impacts. Consistent with previous studies, women and younger individuals reported higher rates of anxiety, depression and post-traumatic stress symptoms following exposure to gun violence9,10. These patterns suggest that early identification and intervention in these groups should be prioritized, with trauma-informed mental health services tailored to their specific needs.

Moreover, community-level exposure to gun violence, even in the absence of direct victimization, was linked to substantial psychological distress. This finding suggests the need for mental health support systems that extend beyond individual victims to entire communities impacted by recurrent gun violence3,4. Public health strategies must include comprehensive community-based interventions that provide resources and mental health care to areas experiencing high rates of firearm violence. Social support systems addressing the mental health needs of entire communities, rather than focusing solely on individual victims, are essential to mitigating the long-term public health burden of gun violence11.

The observed prevalence of community-based mass shootings in this Article (20.1%) reflects a broad definition of ‘community’ and lifetime exposure, distinguishing our measurement strategy from previous studies focusing on single incidents. Notably, findings from ref. 25 indicate that 18.5% of individuals living within a 10-mile radius of a mass shooting reported being on site or knowing someone on site, confirming the widespread reach of mass-shooting events. Our estimates for firearm threats (18.4%) also closely approximate national rates of 21% in previous research16. Furthermore, when comparing lifetime firearm threats across demographics, our results are consistent with those of ref. 17, who reported that 21.7% of Black adults and 30.2% of Native American adults had been threatened with a firearm, aligning with our findings (not reported in tabular form) of 21.0% for Black adults and 34.7% for Native Americans. Reference 17 also observed that 38.2% of Black adults and 27.9% of Native Americans reported witnessing or hearing about an intentional shooting in their neighborhood, a form of community exposure comparable to the broad lifetime exposures assessed in our survey17.

Despite the consistency of our findings with previous research and the robust quality checks outlined in Methods, it is essential to emphasize the inherent limitations of self-reported data. While underreporting of mental health symptoms has been documented—particularly among men, who may be less likely to disclose internalizing symptoms such as depression and anxiety due to social stigma26—overreporting also remains a plausible concern given the high prevalence of self-reported psychological distress in this study. The majority of respondents reporting gun violence exposure indicated some form of psychological impact, with rates far exceeding general population estimates for depression, anxiety and PTSD. Previous research with a random sample of people in six communities that had experienced a major mass shooting found it had persistent and widespread psychological consequences, with nearly a quarter of surveyed adults meeting presumptive diagnostic criteria for past-year PTSD22. Our sample was larger and broader, but our focus on internalizing symptoms may have further limited the scope of reported impacts as externalizing behaviors, such as substance use and aggression, which may be more prevalent among men following trauma27, were not measured. Without objective measures to confirm exposure or clinical diagnoses, the potential for reporting bias cannot be excluded. Future research should incorporate longitudinal designs, external validation and objective indicators of both exposure and mental health outcomes to improve the reliability and validity of findings and expand the understanding of the full spectrum of psychological consequences of gun violence.

The cross-sectional nature of the data also limits the ability to draw causal inferences regarding the long-term effects of gun violence exposure. While respondents self-reported that their mental health was impacted by gun violence, it is not possible to rule out the influence of other life circumstances jointly underpinning mental health and gun violence exposure. As noted elsewhere11, longitudinal studies are needed to better understand the persistence of mental health impacts over time and to identify factors that may contribute to either recovery or chronic psychological distress.

Finally, political orientation emerged as a significant correlate of mental health impacts in this Article. Respondents who identified as Democrats were more likely than Republicans to report mental health effects following exposure to gun violence. These findings are consistent with previous research on political partisanship and psychological outcomes during the COVID-19 pandemic28. Independent voters were most likely to experience the long-term effects of exposure. Further research is needed to disentangle the directionality of this relationship—whether political affiliation influences the perception of mental health impacts, or whether traumatic exposure affects political identity, particularly as firearm violence remains a politically charged issue in the United States.

Conclusion

This study provides insight into the notable and enduring mental health consequences of gun violence exposure in the United States, affecting both individuals directly impacted and entire communities. By highlighting the differential effects of mass and non-mass shootings on mental health outcomes, this research underscores the urgent need for targeted, trauma-informed mental health interventions that address the specific vulnerabilities of women, younger individuals and those exposed to non-mass shootings. Importantly, the findings also emphasize the broader psychological toll of community-level exposure, suggesting that public health strategies must extend beyond individual victim support to include community-based mental health resources. Future research should focus on the long-term trajectories of psychological recovery and the structural factors—such as social support and health-care access—that influence these outcomes. As gun violence remains a pervasive public health crisis in the United States, comprehensive mental health interventions at both the individual and community levels are essential for mitigating its profound societal impact.

Methods

Data

The authors contracted with an online market research firm (YouGov) to field a national survey of a cross-section of adults in the United States from 16 January to 30 January 2024. Guided by findings on mass shootings and victimizations from the Gun Violence Archive29 and the National Crime Victimization Survey30, a sample of 10,000 individuals residing in the United States was generated using multi-stage matched-design sampling (Table 1). YouGov aims to model the population of interest (in our case, US adults) using their online, opt-in panel. The target population was enumerated using a synthetic sampling frame—by age, education, gender and race/ethnicity—based on several national population data sources, including the American Community Survey public use microdata file. Sample matching was conducted by first randomly drawing a target sample and then selecting individuals to be interviewed from their large pool of panelists matched to the target sample. Interviewed respondents were assigned weights using propensity scores derived from a logistic regression model with age, education, gender, region and race/ethnicity to estimate inclusion in the sample frame. Post-stratified weights were then generated on the basis of a four-way interlocking of age, education, gender and race/ethnicity demographic categories and the 2020 presidential vote (including non-voters). All analyses were weighted to produce values designed to be representative of US adults.

Individuals who opted in to YouGov’s panel were sent a generic email inviting them to participate in an online survey. They were prompted to click a link that directed them to a landing page describing the study, ‘Exposure to Gun Violence in the United States’; 13,425 individuals accessed the link. Upon providing informed consent digitally, respondents completed an online self-administered custom survey designed to gather information on exposure to gun violence. The survey followed the American Association for Public Opinion Research guidelines. It included sections on demographics, political ideation, religion, ecological and legal experiences, direct and indirect exposure to mass shootings and gun injury, firearm ownership and vignettes. Respondents completed the survey in approximately 12 min, with the number of questions varying on the basis of their exposure history. After removing individuals as part of screenouts, cleanouts, partial completion, exceeding demographic quotas and matches, YouGov provided us with a dataset containing 10,000 respondents and sample weights designed to be representative of US adults. YouGov compensated participants for survey completion using a point-based system. All data, code and materials used in the analysis are archived on the Open Science Framework.

Measures of gun violence exposure

Six forms of gun violence were assessed, with respondents queried about their lifetime exposure to mass and non-mass shootings. Mass shootings were defined as “gun-related crimes where four or more people are shot in a public space, such as a school, shopping mall, workplace, or place of worship,” which blended the Congressional Research Service’s definition of a ‘mass public shooting’31 and the Gun Violence Archive’s mass-shooting definition29, designed to be inclusive of victims and accessible to the public32.

Three specific measures captured mass-shooting victimization. Community exposure measured mass shootings that occurred “in your local community,” defined as “a geographic area in which you reside or to which you feel especially close, such as a neighborhood, small city or area in a larger city, or place where you spend a large amount of your time, such as a workplace, place of worship, or recreational area” and “while you were living there.” Two measures of direct exposure to mass shootings (direct exposure: present, uninjured; direct exposure: present, injured) were derived from a single prompt, where respondents were asked “Have you personally ever been physically present on the scene of a mass shooting in your lifetime?” The survey clarified that by ‘physically present’ we mean ”in the immediate vicinity of where the shooting occurred at the time it occurred, such that bullets were fired in your direction, you could see the shooter, or you could hear the gunfire.” Respondents who answered ‘yes’ to being physically present on the scene of a mass shooting were asked questions about the incident, including, “Were you physically injured in the incident? (which could include being shot, trampled, or something else that caused physical injury),” a broader operationalization of victimization than relying solely on being directly struck by gunfire.

For ‘exposure to any gun injury beyond mass shootings,’ three specific forms of victimization were assessed: (1) threatened with a gun, (2) shot at but not struck and (3) intentionally shot. All items were coded dichotomously on the basis of whether (yes, 1) or not (no, 0) they occurred in the respondent’s lifetime. The section of the survey concerned personal as well as family gun violence victimization, where respondents were presented with a list of various forms of gun violence and asked “Which of the following, if any, have happened to you?”

If a respondent indicated any experience with gun violence, they were queried on mental health impacts and duration. Each form of gun violence to which a respondent was subject would prompt the question “After the incident, did you experience any of the following?” Respondents could select that they had anxiety/fear, depression (including difficulty sleeping, changes in eating and trouble concentrating), panic attacks and post-traumatic stress symptoms. Each type of mental health impact was coded dichotomously, where ‘yes’ was 1 and ‘no’ was 0. Respondents were also asked, “How long did your experience with [mental health impact] last?” These impacts were coded dichotomously (yes, 1; no, 0). Duration of the mental health impacts was measured with options ranging from days, weeks, months and years, dichotomized into long-term (years, 1) or not (days, weeks and months, 0).

Measures of demographic, economic, political and social correlates

Table 1 provides descriptive statistics for the sociodemographic correlates. Age was measured in years by subtracting birth year from 2024. Male refers to self-identification as a male, coded 1, or female, coded 0. Respondents could self-identify with several racial/ethnic groups, including white, Black, Hispanic, Asian and other/multi, the latter of which included multiple races/ethnicities, Middle Eastern and American Indian/Pacific Islander. White respondents were used as the reference group. Respondent education was based on six response categories, ranging from no high school to post-graduate degree completion, of the highest level of education completed, converted to years. Sixteen categories of annual family income (in thousands), ranging from less than $10,000 to $500,000 or more, were converted to mid-point dollar amounts. Respondents who did not respond to family income were assigned the mean value, and a dummy variable was included in all models. If respondents described themselves as “leaning,” being a “not very strong” or a “strong” Democrat or Republican, they were recorded as such, while those who stated they were independent or “not sure” were recorded as Independent: self-identified Republicans were used as the reference category. News interest, which captures the frequency with which respondents monitor government and public affairs, including hardly at all / don’t know (15.0%), only now and then (15.5%), some of the time (30.2%) and most of the time (39.4%), was used as a control variable; the latter was the focal category and the former three were pooled as the reference category.

Comparisons with probability samples

The values for demographic, economic, political and social variables closely approximate national estimates of adults derived from the 2022 American Community Survey (ACS) 1-year estimates. The sample is 3.9% from the Silent Generation born before 1946 (ACS = 6.6%), 28.1% from the Baby Boomer Generation with birth years between 1946 and 1964 (ACS = 25.4%), 25.3% from Generation X with birth years between 1965 and 1980 (ACS = 25.1%), 27.9% from the Millennial Generation with birth years between 1981 and 1996 (ACS = 27.5%) and 14.8% from Generation Z born 1997 or more recently (ACS = 15.4%). Women are slightly more likely (51.3%) to be represented as respondents than men (48.7%), consistent with ACS estimates (women = 50.9%, men = 49.1%). The predominant racial/ethnic groups in the United States were consistently represented in the sample regarding ACS estimates: 62.8% white (ACS = 63.3%), 16.0% Hispanic (ACS = 17.2%) and 12.5% Black (ACS = 11.9%). Levels of educational attainment and regional representation in the study sample also closely match ACS data. High school diploma or lower was the highest level of education for 39.4% of respondents (ACS = 37.7%), 27.9% completed some college but fell short of a baccalaureate degree (ACS = 29.3%), 21.5% completed a 4-year degree (ACS = 20.6%), and 12.3% completed a graduate degree (ACS = 12.4%). Income categories could not be directly compared because of YouGov’s categorization of family income.

Whereas YouGov’s matched sample design induces similarity in sociodemographic and other characteristics, alternative indicators may produce different results from probability samples derived from random digit dialing and housing units. Therefore, it is important to assess the generalizability of the inferences reached in this Article by comparing findings with established probability samples. We used several alternative indicators as substantively relevant generalizability checks.

Fear of victimization is regularly asked as part of Gallup polling and the General Social Survey. An identical item was administered to respondents in this YouGov sample, which read as follows: “Is there any area near where you live—that is, within a mile—where you would be afraid to walk alone at night?” The results were consistent across samples. In this YouGov sample, 39.8% of respondents said yes. Gallup administered this question in October 2021 and October 2023, where 37% and 40% of respondents answered in the affirmative33. The General Social Survey administered this question in 2022, where 37% of respondents stated yes34.

Arrest and incarceration could also be compared with estimates derived from probability samples. In our survey, 24.0% (95% CI: 23.1–24.9%) self-reported arrested in their lifetime. Estimates of arrest are 30% for respondents in the Add Health survey by ages 24 to 34 (ref. 35) and 30% for respondents in the National Longitudinal Survey of Youth 1997 by age 23 (ref. 32); we expected our estimates to be higher owing to criminal justice system expansion and including older generations in our sample36. Indeed, when we restrict our sample to the National Longitudinal Survey of Youth 1997 birth cohorts (1980–1984), we obtain lifetime arrest estimates of 29.4% (CI: 26.1–32.7%). Lifetime estimates of incarceration in a county jail or federal or state prison were 13.5% (95% CI: 12.7–14.2%). Estimates of incarceration vary widely in previous research, depending on birth cohort and measurement, ranging from 6% to 19%37,38,39,40,41, but our findings most closely approximate the lifetime incarceration estimate of 11.4% from the National Epidemiological Survey on Alcohol and Related Conditions-III, a probability sample of 36,309 adults39.

Overall, these findings suggest YouGov’s matched sample design can yield generalizable inferences.

Quality checks and non-random selection

Online, opt-in panels offer many advantages to researchers, including flexibility, timeliness and cost-effectiveness. They are not immune to concerns about generating population and relational inferences. A large N sample, combined with multi-stage matched-design sampling, minimizes some of these concerns. Still, it is important to undergo quality checks beyond comparisons with probability samples and examine non-random selection in a custom survey that aims to generate information on the mental health impacts of exposure to mass shootings.

Quality checks were undertaken by examining responses to supplemental questions about local and personal exposure to shootings. Inquiries into the place where personal exposure to a mass-shooting incident occurred and the reason for presence at the location resulted in removing a small number of incidents in which respondents were not present on the scene (n = 4), cases in which incidents occurred in war zones as part of military service (n = 16) and cases that occurred outside of the United States that appeared independent of military service (n = 20). There were also six instances where local community exposure occurred outside the United States. This was in addition to quality checks undertaken by YouGov to screen out respondents who did not meet quality control standards (for example, speed throughs, attention checks, non-sensical responses and duplicate internet protocol addresses).

To address the prospect of non-random selection into a custom survey titled ‘Exposure to Gun Violence in the United States,’ we determined whether similar results could be obtained using an omnibus survey. An omnibus survey is comparable to a custom survey because it is generated using the same sampling strategy; it differs in that a wide range of questions are included, rendering respondents unaware of the content upon agreeing to complete a survey. We inserted the personal exposure to a mass shooting question into an omnibus survey fielded the same month as the custom survey to 3,000 respondents, finding that 7.77% of respondents indicated that they were present on the scene of a mass shooting. This estimate closely approximates the observation from the custom survey (4.8% present, uninjured + 2.2% present, injured = 7.0%), which combined with being derived from a separate sample enhances confidence in the findings15.

Analytic plan

Descriptive statistics were computed for each form of gun violence and the corresponding mental health impacts. Bivariable ordinary least squares regression was used to determine differences in the prevalence of any/long-term mental health impacts between each of the six types of gun violence among the exposed subsample. Multivariable logistic regression was used to generate adjusted odds ratios that estimated the associations between sociodemographic characteristics and the presence and (long-term) duration of mental health impacts. All analyses were weighted to provide nationally representative estimates and report robust standard errors and were conducted in Stata 18.0.

Ethics statement

The institutional review board at Hamline University approved the study protocol (no. 2023-11-267ET).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.