Political Violence In The U.S.

In recent years, the United States has seen an increase in the number of left-wing terrorism attacks and plots, although such violence has risen from very low levels and remains much lower than historical levels of violence carried out by right-wing and jihadist attackers. So far, 2025 marks the first time in more than 30 years that left-wing terrorist attacks outnumber those from the violent far right. Despite its decline this year, right-wing terrorism could easily return to previous high levels. It is important to resource counterterrorism efforts against both right- and left-wing terrorism and work with communities to gain their support against extremists. Leaders across the political spectrum must condemn violent extremism of all kinds, denying it legitimacy and reducing its appeal.

Introduction

The tragic killing of political commentator and conservative activist Charlie Kirk has once again put the spotlight on political violence in the United States, with figures on both sides of the political aisle decrying extremism on the other. To understand the danger of political violence today and to find the best solutions for reducing it, it is important to understand the overall threat landscape and how both left- and right-wing violence have evolved and could change in the future.

Our analysis of terrorism trends in the United States shows that, indeed, left-wing violence has risen in the last 10 years, particularly since President Donald Trump’s rise to political prominence in 2016, although it has risen from very low levels and remains much lower than historical levels of violence carried out by right-wing and jihadist attackers. More contentious politics in the United States and the expansion of the Make America Great Again (MAGA) movement appear to have reenergized violent left-wing extremists. The left-wing movement as a whole has not returned to its violent heights of the 1960s and 1970s, but the number of terrorist incidents involving left-wing extremists so far this year puts 2025 on pace to be the left’s most violent year in more than three decades. Moreover, 2025 marks the first time in more than 30 years that left-wing attacks outnumber those from the far right.

Indeed, the increase in left-wing attacks is particularly noticeable because attacks from right-wing perpetrators have sharply declined in 2025. This decline is striking, and explanations are speculative. One possibility is that many traditional grievances that violent right-wing extremists have espoused in the past—opposition to abortion, hostility to immigration, and suspicions of government agencies, among others—are now embraced by President Trump and his administration.

Similarly, jihadist attacks have declined in frequency since their peak in the 2010s, owing largely to the destruction of major groups like al Qaeda and the Islamic State and the subsequent decline in the power of the jihadist ideology to inspire attackers.

Left-wing attacks are remarkably less lethal overall than jihadist or right-wing attacks. However, even incidents that do not result in mass casualties can still have significant impact. Fortunately, many left-wing attackers (though not all) have demonstrated limited skill in carrying out violence, and the movement is disorganized, with little formal coordination.

The rise in left-wing attacks merits increased attention, but the fall in right-wing attacks is probably temporary, and it too requires a government response. In any case, many of the prescriptions for fighting terrorism effectively apply to violence from both the left and right. These include ensuring proper counterterrorism resourcing, avoiding overreactions, and having leadership unequivocally condemn such attacks.

The remainder of this brief is divided into six sections. First, terms such as “left-wing” and “terrorism” are defined. Second, trends in left-wing terrorism in the United States are analyzed, with an emphasis on the increase in the number of incidents since 2016. Third, the causes of the rise of left-wing incidents are assessed. Fourth, weaknesses that limit the impact of left-wing terrorism are examined. Fifth, possible reasons for the decline in right-wing and jihadist terrorism are discussed. The sixth and final section discusses several policy implications that can help combat violence from perpetrators across the political spectrum.

Definitions

This analysis defines terrorism as the deliberate use or threat of premeditated violence by nonstate actors with the intent to achieve political goals by creating a broad psychological impact. Using this definition, CSIS researchers compiled and analyzed a dataset of 750 terrorist attacks and plots in the United States between January 1, 1994, and July 4, 2025. The dataset includes information such as incident date, location, target and location type, weapon used, and victim fatalities, as well as perpetrator age, sex, ideology, group affiliation, and current or former affiliation with the military or law enforcement. A full methodology and codebook for the dataset is available at CSIS.org.

This brief defines left-wing terrorism as that which is motivated by an opposition to capitalism, imperialism, or colonialism; black nationalism; support for LGBTQ+ rights; support for environmental causes or animal rights; adherence to pro-communist, pro-socialist beliefs or “anti-fascist” rhetoric; opposition to government authority under the belief it is a tool of oppression responsible for social injustices; support for decentralized political and social systems, such as anarchism; or partisan extremism, where violence is justified against political opponents and parties perceived as advancing right-wing agendas.

Right-wing terrorism as used in this analysis includes incidents motivated by ideas of racial or ethnic supremacy; opposition to government authority, believing it is tyrannical and illegitimate; misogyny, including incels; hatred based on sexuality or gender identity; belief in the QAnon conspiracy theory; opposition to abortion; or partisan extremism, where violence is justified against political opponents and parties perceived as advancing left-wing agendas.

Note that terms such as “left-wing terrorism” and “right-wing terrorism” as used in this brief do not correspond to mainstream political parties in the United States, such as the Democratic and Republican parties, nor do they correspond to the overwhelming majority of political liberals and conservatives in the United States.

In many cases, clear ideological categorization of perpetrators is difficult. Former FBI Director Christopher Wray once referred to a “salad bar of ideologies,” where perpetrators of violence choose among an array of causes, many of which do not align with a traditional right-left dichotomy or other easy ideological classification.1 In other cases, perpetrators may not have political motives despite their targets. For example, Thomas Crooks, who tried to assassinate then-candidate Donald Trump in Butler, Pennsylvania, in July 2024, reportedly searched online before his attack for locations where both Trump and then-President Joe Biden would be publicly speaking. While it is possible Crooks had political motives, FBI reports and journalist investigations suggest the explanation was more likely a mix of personal issues.

Definitions

This analysis defines terrorism as the deliberate use or threat of premeditated violence by nonstate actors with the intent to achieve political goals by creating a broad psychological impact. Using this definition, CSIS researchers compiled and analyzed a dataset of 750 terrorist attacks and plots in the United States between January 1, 1994, and July 4, 2025. The dataset includes information such as incident date, location, target and location type, weapon used, and victim fatalities, as well as perpetrator age, sex, ideology, group affiliation, and current or former affiliation with the military or law enforcement. A full methodology and codebook for the dataset is available at CSIS.org.

This brief defines left-wing terrorism as that which is motivated by an opposition to capitalism, imperialism, or colonialism; black nationalism; support for LGBTQ+ rights; support for environmental causes or animal rights; adherence to pro-communist, pro-socialist beliefs or “anti-fascist” rhetoric; opposition to government authority under the belief it is a tool of oppression responsible for social injustices; support for decentralized political and social systems, such as anarchism; or partisan extremism, where violence is justified against political opponents and parties perceived as advancing right-wing agendas.

Right-wing terrorism as used in this analysis includes incidents motivated by ideas of racial or ethnic supremacy; opposition to government authority, believing it is tyrannical and illegitimate; misogyny, including incels; hatred based on sexuality or gender identity; belief in the QAnon conspiracy theory; opposition to abortion; or partisan extremism, where violence is justified against political opponents and parties perceived as advancing left-wing agendas.

Note that terms such as “left-wing terrorism” and “right-wing terrorism” as used in this brief do not correspond to mainstream political parties in the United States, such as the Democratic and Republican parties, nor do they correspond to the overwhelming majority of political liberals and conservatives in the United States.

In many cases, clear ideological categorization of perpetrators is difficult. Former FBI Director Christopher Wray once referred to a “salad bar of ideologies,” where perpetrators of violence choose among an array of causes, many of which do not align with a traditional right-left dichotomy or other easy ideological classification. In other cases, perpetrators may not have political motives despite their targets. For example, Thomas Crooks, who tried to assassinate then-candidate Donald Trump in Butler, Pennsylvania, in July 2024, reportedly searched online before his attack for locations where both Trump and then-President Joe Biden would be publicly speaking. While it is possible Crooks had political motives, FBI reports and journalist investigations suggest the explanation was more likely a mix of personal issues.

Left-Wing Terrorism Incidents Are on the Rise

The first half of 2025 was marked by an increase in left-wing terrorist attacks and plots in the United States, which continues a trend noticeable over the last decade. In absolute terms, left-wing incidents are on track in 2025 to reach historically high levels in the last 30 years, as shown in Figure 1.

From 1994 through 2000, there was an average of 0.6 left-wing incidents annually; in the following decade, that figure doubled to 1.3 a year. Numbers began to grow substantially, however, in 2016, and from 2016 to 2024, they averaged 4.0 a year. Through July 4, 2025 (thus excluding the Kirk attack), there were five left-wing attacks or plots, which sets a trajectory for a record-breaking year in the last 30 years.

▲ Figure 1: Left-Wing Terrorist Attacks and Plots in the United States, 1994–2025. Source: Data compiled by CSIS Warfare, Irregular Threats, and Terrorism Program.

This elevated number of left-wing incidents is even more striking when compared with the number of incidents classified under other ideological orientations. Left-wing terrorist attacks and plots as a percentage of all terrorist attacks and plots were at a record high in the first half 2025, although the decline of other forms of terrorism plays a significant role in this relative increase.

▲ Figure 2: Left-Wing Terrorist Attacks and Plots as a Percentage of All Attacks and Plots, 1994–2025. Source: Data compiled by CSIS Warfare, Irregular Threats, and Terrorism Program.

▲ Figure 3: Terrorist Attacks and Plots in the United States by Perpetrator Orientation, 1994–2025. Source: Data compiled by CSIS Warfare, Irregular Threats, and Terrorism Program.

Indeed, a dramatic decline in right-wing incidents in 2025 has contributed significantly to the relative increase in left-wing incidents. So far, 2025 is the first year in the CSIS data where the number of left-wing incidents is greater than the number of right-wing ones.

▲ Figure 4: Left-Wing vs. Right-Wing Terrorist Attacks and Plots in the United States, 1994–2025. Source: Data compiled by CSIS Warfare, Irregular Threats, and Terrorism Program.

Though the number of left-wing terrorist attacks and plots is experiencing a rise, the effectiveness of perpetrators typically remains limited. Although left-wing perpetrators often carry out their plans, they rarely do so with deadly effect. Two metrics illustrate this dynamic.

First, a large share of left-wing plots succeed in becoming actual attacks. In 2025, of five left-wing incidents that occurred before July 4, four were attacks carried out and only one was a disrupted plot. This continues a long-standing pattern. As shown in Figure 5, the number of left-wing attacks carried out is typically much greater than the number of plots disrupted. This trend is mirrored in right-wing and ethnonationalist incidents and is likely influenced by bias in data collection. That is, incomplete public information means disrupted plots are likely undercounted.

▲ Figure 5: Disrupted Plots vs. Attacks Among Left-Wing Terrorist Incidents in the United States, 1994–2025. Source: Data compiled by CSIS Warfare, Irregular Threats, and Terrorism Program.

By contrast, jihadist incidents exhibit the opposite pattern. Disrupted plots are far more common than successful attacks, probably a reflection of decades of intense intelligence and law enforcement focus on jihadist activity after 9/11, as well as the far greater media publicity given to disrupted jihadist attacks, which enables greater data collection.

▲ Figure 6: Disrupted Plots vs. Attacks Among Jihadist Terrorist Incidents in the United States, 1994–2025. Source: Data compiled by CSIS Warfare, Irregular Threats, and Terrorism Program.

The fact that left-wing plots so often result in completed attacks elevates the significance of the recent rise in incidents because it indicates that the recent increase is likely to translate into realized violence.

Second, despite the rise in the number of left-wing incidents and the likelihood that such incidents involve realized violence, the lethality of left-wing attacks remains very low. Left-wing attacks are overwhelmingly non-lethal and far less lethal compared with other ideological orientations. Since 2020, only two fatalities have resulted from left-wing terrorist attacks in the United States: Luigi Mangione’s assassination of UnitedHealthcare CEO Brian Thompson in New York City in December 2024 and Michael Reinoehl’s fatal shooting of right-wing protester Aaron Danielson in Portland, Oregon, in August 2020 (if the Kirk killing is included, as seems likely, it would be a third fatality). Right-wing and jihadist attacks, by contrast, have caused far higher fatalities.

In the past decade, despite the increase in the number of left-wing incidents, left-wing attacks have killed 13 victims, compared with 112 and 82 victims for right-wing and jihadist attacks, respectively. Some of the key factors driving these dramatic discrepancies are explored in a later section of this brief.

▲ Table 1: Victims Killed in Terrorist Attacks by Perpetrator Orientation, 2016–2025. Source: Data compiled by CSIS Warfare, Irregular Threats, and Terrorism Program.

What’s Causing the Rise in Left-Wing Incidents?

The increase in left-wing incidents in the past decade is driven by plots and attacks directed at government and law enforcement targets. Of the 41 left-wing incidents since 2016, anti-government extremism motivated 17 of them, and partisan extremism motivated another 11. All left-wing attacks through July 4, 2025, appeared to be motivated by one of these ideologies, and the Kirk killing fits this pattern, although details about Kirk’s alleged killer are still emerging.

The only significant break from this trend was a surge of six left-wing firebombings against pro-life targets (pregnancy crisis centers and the office of an anti-abortion group) in the summer of 2022 around the time of the overturning of Roe v. Wade. These attacks were intentionally perpetrated at night against unoccupied buildings to reduce (though not eliminate) the risk to people.

To understand rising left-wing violence, it is useful to distinguish between partisan extremism and anti-government extremism.

Partisan extremism includes attacks and plots against elected officials, political candidates, political party officials, and political staff and workers from terrorists with opposing political views. For example, on January 28, 2025, U.S. Capitol Police arrested Riley Jane English, a 24-year-old from Massachusetts, on the National Mall in Washington, D.C., after she approached officers and revealed she was carrying a folding knife, two Molotov cocktails, and a lighter. According to prosecutors, English said she intended to kill senior U.S. officials, initially identifying Defense Secretary Pete Hegseth, whom she described as a “Nazi,” before shifting her focus to House Speaker Mike Johnson and then to Treasury Secretary Scott Bessent. She also expressed a desire to attack the conservative Heritage Foundation.

In a separate incident on March 30, 2025, an assailant set fire to the headquarters of the Republican Party of New Mexico in Albuquerque, igniting the entrance late at night. Graffiti reading “ICE = KKK” was spray-painted on a wall near the site.

Attacks such as English’s plot to assassinate senior officials and the arson of the Republican Party’s New Mexico headquarters reflects the most severe effects of polarization in the United States. A variety of survey data underscores that this issue is widespread and worsening. In 2016, fewer than half of Republicans or Democrats described the opposing side as immoral, dishonest, or unintelligent. By 2022, however, most Republicans described Democrats as immoral (72 percent), dishonest (63 percent), and unintelligent (70 percent), while most Democrats said the same of Republicans (64 percent, 61 percent, and 52 percent, respectively).

Similarly, while less than 4 percent of Americans express support for partisan violence such as assault, arson, or murder, both sides greatly overestimate their opponent’s willingness to endorse such actions, with Democrats believing 45.5 percent of Republicans support partisan murder, and Republicans believing that 42 percent of Democrats do.

Although the vast majority of Americans would never commit partisan violence and oppose it, widespread polarization and misperceptions that the other side is far more violent than it actually is creates a dangerous environment where extremists can more easily rationalize using violence. Growth in even a tiny minority who are willing to commit partisan violence has the potential for tremendous consequences considering the combustible political climate in the United States and the fact that symbolic and strategically important political leaders are among the potential targets.

In addition to partisan extremism, anti-government extremism has also become more pronounced as a motive for left-wing attacks in 2025, particularly around the issue of immigration.

In one incident on July 4, 2025, a group attacked the U.S. Immigration and Customs Enforcement (ICE) Prairieland Detention Facility in Alvarado, Texas. According to federal prosecutors and law enforcement statements, approximately a dozen individuals dressed in black clothing and equipped with tactical gear initiated the attack by detonating fireworks and spray-painting anti-ICE graffiti on vehicles and structures outside the facility. When an Alvarado police officer responded to the scene, an assailant positioned in a nearby wooded area opened fire, wounding the officer in the neck. Meanwhile, another individual reportedly fired 20 to 30 live rounds at unarmed ICE correctional officers outside the facility. Law enforcement subsequently apprehended 14 suspects, who now face federal charges including attempted murder of federal officers and firearm-related offenses. Searches uncovered AR-style rifles, a pistol, body armor, two-way radios, spray paint, and flyers with anti-ICE slogans. One of the alleged attackers stored cellphones inside Faraday bags, which are used to block signals and indicate premeditated efforts to evade law enforcement tracking.

Traditionally, anti-government extremism refers to violence aimed at state institutions viewed as illegitimate or oppressive, whereas partisan extremism targets specific political figures or individuals based on party affiliation. As events such as the Prairieland ICE attack suggest, left-wing opposition to the Trump administration in 2025 often manifests as both: rejecting its political leadership and resisting its efforts to expand the authority of military and law enforcement institutions. Together, partisan and anti-government extremism help explain why 2025 has seen an escalation of left-wing violence, as opposition to the Trump administration fuels attacks against both its political leadership and the state institutions that carry out its agenda.

Left-Wing Weaknesses

Despite the rise in the number of left-wing terrorist incidents, there exist several characteristics and conditions that limit the scale and sophistication of attacks. The overall low lethality rates in left-wing attacks are probably attributable to several factors, including target selection, target scope, tactical methods employed, low levels of perpetrator skill, and counterterrorism measures.

First, left-wing perpetrators typically select targets that limit opportunities for mass killing. In the past decade, left-wing attacks most commonly occurred at government or law enforcement facilities. These locations are often protected by physical fortifications and security personnel, making it more difficult for perpetrators to kill targets during an attack. For example, in the July 2025 attack on ICE’s Prairieland Detention Facility, one police officer was wounded before additional law enforcement forced the attackers to flee. The attackers’ choice of a hardened federal compound with security personnel on site contributed to the attack’s failure to kill any victims. This contrasts with jihadist attacks, which most commonly target crowded public areas with limited or no security measures, such as the car ramming attack on Bourbon Street in New Orleans on January 1, 2025, that killed 14 victims.

Second, left-wing violence is often narrowly directed at specific individuals rather than indiscriminate killings of civilians. Most notably, 10 of the 13 victim fatalities from left-wing attacks in the past decade have been police officers ambushed in public areas by attackers using firearms. This pattern underscores that even the deadliest left-wing attacks have tended to focus on targeted confrontations with law enforcement rather than mass-casualty events.

Third, left-wing perpetrators frequently employ tactical methods poorly suited to producing mass casualties. In the past decade, 20 of 35 left-wing attacks have involved the use of incendiaries or arson as the primary weapon. Incendiaries and acts of arson typically lack precision, resulting in damage that can be severe but not necessarily lethal. In some cases, the arson was done at night, which further reduced the likelihood of fatalities because the targets were often sites that were largely unoccupied.

Taken together, the typical target selection, scope, and weapon selection of left-wing attackers reflect an intent to signal opposition or cause disruption rather than inflict mass casualties.

Fourth, left-wing perpetrators frequently lack the weapons and tactical training to maximize their impact. For example, on February 14, 2022, Quintez Brown, a 21-year-old black nationalist, entered the Louisville campaign office of Democratic mayoral candidate Craig Greenberg with a 9 mm pistol. Prosecutors argued that Brown saw himself as an “equalizer” striking at a symbol of gentrification and oppression. Despite firing multiple rounds at point-blank range, Brown missed his target, with one bullet only grazing Greenberg’s sweater.

Fifth, unlike many foreign terrorist organizations with centralized leadership, funding, and training infrastructures, left-wing terrorists, like most terrorist actors in the United States, operate as loosely affiliated networks or as lone individuals, limiting their ability to train and to plan and execute complex operations. Social and technological factors, including online radicalization, often result in isolated actors lacking the resources, expertise, or coordination needed for sophisticated attacks.

The lack of organization also creates a multiplicity of competing goals that hinders strategic effectiveness. Like their right-wing counterparts, left-wing terrorists are against many things, and there is no clear prioritization of targets within the movement. Similarly, they are unable to calibrate violence, making it more likely to backfire.

Sixth and finally, U.S. law enforcement and intelligence agencies have developed robust counterterrorism measures, particularly since 9/11, that disrupt plots and largely deter large-scale attack planning across all ideologies. Legal restrictions, such as controls on explosives and surveillance of known extremist networks, further constrain terrorist operational capacity. Social media, in addition to enabling radicalization, also hinders operational security, revealing intentions and possible actions of individuals who, in the past, might have remained undetected. Combined, these dynamics help ensure that, while the threat of terrorism remains, the capacity of all U.S.-based terrorist movements to execute attacks is significantly diminished.

Why Have Jihadists and Right-Wing Incidents Fallen?

Explaining non-events, such as why attacks from rival ideological groups and individuals have fallen, is always difficult, but there are several possible reasons for the decline in jihadist and right-wing terrorism.

For jihadists, the main foreign terrorist groups in recent years—al Qaeda and the Islamic State—are far weaker than they were at their peaks. The United States and its allies have killed group leaders, often leaving them in disarray. Al Qaeda, for example, still has not named a successor since the death of Ayman al-Zawahiri in 2022. The Islamic State likewise has lost numerous leaders, and it no longer has an above-ground caliphate where it can train people. In both cases, the losses have made the group less inspiring, although some bottom-up radicalization remains a concern. Finally, factors like aggressive law enforcement and a U.S. Muslim community that collaborates with law enforcement lead many would-be attackers to be caught in early stages of plotting.

The sudden decline in right-wing terrorism is both more striking and harder to explain. From 1994 through 2000, there was an average of 21 right-wing attacks or plots each year. In the following decade, right-wing incidents fell to an average of 7 annually. From 2011 through 2024, right-wing incidents climbed back up to an average of 20 a year. In the first half of 2025, however, there was only one right-wing terrorist incident in the United States—the killing of Minnesota state legislator Melissa Hortman and her husband in June—a remarkable drop off.

Aggressive law enforcement efforts under former President Biden hurt the right-wing extremist movement, with the January 6 investigation in particular causing disarray. The U.S. government brought charges against over 1,000 individuals, including many leaders of groups like the Oath Keepers and Proud Boys. In addition, like left-wing terrorism, right-wing terrorism is highly decentralized, and the vast majority of the most lethal attacks in the last 10 years have been perpetrated by lone actors linked to various networks but not tied to any group.

Trump’s election, however, appears to have changed the threat. Although it is impossible to definitively prove the link between the policies of and positions championed in Trump’s second term and the decline in right-wing terrorism incidents in the United States, it is probable that at least some extremists do not feel the need to act violently if their concerns are being addressed.

Most notably, the administration has aggressively targeted immigrants, with high-profile efforts to identify, detain, and deport them. Anti-immigrant sentiment is one of the most important violent extremist motivations in recent years. The Trump administration has also warned of “deep state” abuses, criticized and abolished programs involving diversity, promoted some conspiracy theories, and hired individuals who openly embraced white supremacy. In addition, Trump’s victory temporarily ended many concerns about a Democrat-orchestrated “stolen election,” a leading conspiracy that motivated many extremists in the past. Enrique Tarrio, the former Proud Boys leader and a convicted seditionist whom Trump pardoned, recently summed up the president’s potential psychological effect on the violent far right: “Honestly, what do we have to complain about these days?”

At the same time, it is important to stress that correlation does not mean direct causation. The administration’s rhetoric and policies may overlap with themes found in extremist discourse, but that does not necessarily imply intent to encourage violence. Rather, extremist actors may perceive mainstream political validation of their grievances as reducing the need for independent mobilization or, more concerningly, as tacit endorsement of their worldview in situations where the administration faces resistance or does go far enough in the eyes of an extremist.

What Is to Be Done?

The best responses to the Kirk killing and political violence from any source involve few theatrics but can be highly effective.

First, the government should avoid overreacting with crackdowns on peaceful organizations, which will serve to strengthen extremist views. Radicals will argue that peaceful politics will inevitably fail and that only violence will make a difference. In response to the Kirk killing, President Trump warned that a “radical left group of lunatics” are engaged in a campaign of violence. Other conservative voices, from members of Congress to online influencers, have similarly claimed that the left is engaged in “war.” Kirk’s shooter appears to have acted alone, but Trump has claimed that a network of political organizations fund and support violence, and must be neutralized—a threat that, if acted on, could lead to government action against an array of non-violent organizations whose political positions were anti-Kirk and are anti-Trump. These actions will be counterproductive for combatting terrorist threats.

Second, it is important to resource all dimensions of the terrorism threat. Left-wing terrorism is a Trump administration priority, but jihadist terrorism also remains a concern even though it has declined. Right-wing terrorism could come roaring back, especially if in 2028 there are complaints of a “stolen election” or similar incendiary claims. Developing the programs and expertise to suppress different forms of terrorism takes years, and ignoring a long-term threat to go after a more immediate one could be deadly over time.

Finally, although leaders are not responsible for extremists in their midst, they are responsible for how they behave toward extremists. U.S. political leaders and activists need to lead by condemning violence on their side and calling for calm when it involves violence on the other side. The American Muslim response to jihadist terrorism offers a useful model. Muslim leaders came together to repeatedly condemn jihadist violence, and this reduces the appeal of terrorism. When the mainstream condemns an attack, the individual is less likely to be seen, and see themselves, as a hero or successful agitator, and the community as a whole is more likely to work with law enforcement.

Many leading Democrats have vehemently condemned the Kirk shooting. For their part, many prominent Republicans also immediately condemned right-wing attacks in recent years, including the assassination of Hortman earlier this year and the attack on Paul Pelosi in 2022. But the track record is far from perfect. The celebrations among some on the left of Luigi Mangione is a failure to undermine support for left-wing violence. Similarly, the failure of some conservative leaders to condemn white supremacists and other violent extremists is a major problem, allowing these extremists to believe they are carrying out the will of a broader political movement.

Utah Governor Spencer Cox has served as a model in unequivocally denouncing extremists. Cox described Kirk’s killing as “an attack on all of us.” And he offered a simple exhortation that would benefit both sides, particularly in moments like these, when violence can spiral: “Disagree better.”

APPENDIX

WHAT IS EXCLUDED?

This analysis excludes a number of incidents that further contribute to the perception of an increase in left-wing violent activity in 2025. Different definitions and coding might legitimately include these, but the explanation below details why they are excluded from this dataset.

The definition of terrorism used here excludes the series of attacks against Tesla vehicles and facilities. There were more than 20 such incidents in the United States from January to April 2025. Attacks on Tesla in the United States were linked to individuals expressing opposition to CEO Elon Musk’s political affiliations, particularly his role and actions in the Trump administration. Although the CSIS study team determined these attacks were incidents of economic vandalism rather than terrorism, many involved substantial property destruction and drew sharp condemnation from the Trump administration and the Department of Justice. In some cases, prosecutors have even sought terrorism enhancements in charging decisions. Although excluded from this dataset, these highly publicized incidents attracted significant attention and reinforced the perception of escalating left-wing violence in 2025.

There were three high-profile terrorist attacks in the United States in the first half of 2025 motivated by the conflict between Israel and Palestine. These include the April arson attack on Pennsylvania Governor Josh Shapiro’s residence, a May shooting that killed two Israeli Embassy staff in Washington, D.C., and a June firebombing of a pro-Israel solidarity walk in Boulder that injured 15 demonstrators. In this dataset, these attacks are classified as ethnonationalist incidents, rather than left-wing ones. However, it is noteworthy that all three incidents involved attackers motivated by opposition to Israel’s actions in Gaza or U.S. support for Israel—a political position traditionally (though not always) associated with left-wing politics in the United States.

Finally, political demonstrations against immigration and customs enforcement activity across the United States in 2025 also resulted in many violent encounters, namely between law enforcement and demonstrators. However, most of these incidents did not reach a level of violence that satisfied this study’s definition of terrorism, and they were not intended to cause a broad psychological effect. Nonetheless, they too contributed to the perception of growing left-wing violence.

METHODOLOGY

Data Selection and Process

This research effort aims to identify the scope, nature, and ideology of the terrorism threat in the United States. To analyze this, the CSIS Warfare, Irregular Threats, and Terrorism Program (WITT) compiled a dataset of 750 terrorist attacks and plots that occurred in the United States between January 1, 1994, and July 4, 2025.

This time period was selected in order to provide context on the history of domestic terrorism in the United States in recent decades. The dataset begins in 1994 rather than earlier in the decade due to sourcing challenges. WITT relied heavily on START Global Terrorism Database (GTD) data for the early years of the dataset; a full record of incidents in 1993 is unavailable due to data loss.

WITT drew from the following databases and sources: the Armed Conflict Location & Event Data Project (ACLED); the Anti-Defamation League (ADL)’s Hate, Extremism, Anti-Semitism, and Terrorism (H.E.A.T.) Map; Janes Terrorism and Insurgency Events; START GTD; the Southern Poverty Law Center (SPLC)’s Hate Map; New America’s “Terrorism in America After 9/11” report; and press releases and reports from the Federal Bureau of Investigation (FBI) and Department of Justice (DOJ). WITT cross-referenced events against criminal complaints and affidavits when possible as well as local and national news sources such as the New York Times, Washington Post, and Los Angeles Times.

Incidents were compiled, coded, and reviewed for accuracy and consistency by a team of WITT researchers. External reviewers also routinely review the dataset.

Definition of Terrorism

WITT defined terrorism as the deliberate use or threat of premeditated violence by nonstate actors with the intent to achieve political goals by creating a broad psychological impact of fear or intimidation. For inclusion in the dataset, incidents had to meet all parts of this definition.

Determinations were based on publicly available information as of July 2024.

Common types of events excluded include:

• Economic sabotage: Attacks against infrastructure intended to impose costs on a business or government are often used by terrorist groups and are sometimes considered a form of “economic terrorism.” These attacks were generally excluded from the CSIS dataset for a combination of two reasons: the perpetrator sought to avoid killing or maiming people or the perpetrator intended to send a message to a narrow audience, most frequently the management of a targeted business firm. In practice, this led to the removal of several attacks by environmental groups in the late 1990s and early 2000s. Several of these groups explicitly stated their opposition to harming people and designed their attacks to minimize risk to human life, typically engaging in arson at unoccupied locations at the time of the attack. For similar reasons, many attacks on Tesla facilities and vehicles that occurred in early 2025 were excluded as acts of economic vandalism.

• Escalation at demonstrations: Violence at demonstrations has dramatically increased in recent years. However, these incidents were generally excluded from the CSIS dataset for two main reasons: lack of lethal force and lack of evidence of perpetrators’ desire to have a widespread psychological impact or achieve political goals through violence. Attacks at demonstrations that did meet these criteria were kept in the dataset, such as the Charlottesville car attack in August 2017, where James Alex Fields Jr. deliberately drove his car into a crowd of people peacefully protesting the nearby Unite the Right rally, killing one person and injuring 35. Fields sought to kill counterprotesters indiscriminately and made comments to his mother and posts on social media that suggested premeditation and a desire to spread fear.

• Hate crimes: There is an overlap between terrorism and hate crimes. However, not all hate crimes—defined by the FBI as “crimes in which the perpetrators acted based on a bias against the victim’s race, color, religion, or national origin”—include violent elements. WITT excluded incidents in which there was no actual or serious threat of violence against the physical well-being of people, such as graffiti, trespassing, or damage to objects. WITT also excluded incidents where a hate crime was unlikely to reach a broader audience. In practice, this meant asking whether a reasonable person could expect news of the violence to reach beyond those targeted to a broader political community. Hate crime attacks that were excluded almost always targeted a particular individual based on their membership of a larger community, as opposed to most (though not all) terrorist attacks that seek to kill indiscriminately in order to maximize fear.

• School shootings: Attackers who commit school shootings are typically motivated by a mix of personal grievances, including grievances with classmates, school staff, and romantic partners, as well as a desire to kill, commit suicide, or achieve fame or notoriety. In most school shootings, the attacker does not have clear political objectives, and therefore these incidents are generally excluded from the dataset.

• Support for foreign terrorist organizations: Incidents in which U.S. residents attempted to provide material support to foreign terrorist organizations without connection to specific violent plots or attacks in the U.S. homeland were excluded.

• Attacks involving non-lethal weapons: Attacks using nonlethal weapons were generally excluded, except for attacks clearly intended to kill or maim and cases of arson in contexts where a political issue’s history gave the attack the character of an implicit lethal threat. Most of the latter cases were arson attacks against religious institutions, abortion-related targets, and political offices. Likewise, incidents where perpetrators made deliberate efforts to avoid harming people were excluded, such as instances of economic sabotage or attacks not intended to cause or threaten harm to the physical well-being of people.

• Riots: While riots may involve violence that threatens to kill or maim others, they are often spontaneous, driven by collective unrest, anger, and opportunistic criminal activity that does not target human beings. In many cases, rioters lack the strategic intent to achieve broader political objectives. However, some riots, such as the attack on the U.S. Capitol on January 6, 2021, are included in the CSIS dataset if they involved actors who clearly premediated their violence and had clear political objectives.

• Robberies: Robberies—even those targeting individuals on the basis of ethnicity, race, or religion—were generally excluded from the dataset since the primary motivations of perpetrators are financial, not political. Robberies committed to fund terrorist activities were also excluded, although such activities would be coded separately if they involved a specific plot or attack.

Limitations

There are several potential limitations to the dataset.

First, since WITT drew from multiple data sources—few of which covered most or all of the time period—there is likely inconsistency in how thoroughly events were recorded across years. In particular, fewer sources (including news reports to cross-reference) exist for cases earlier in the time period. In addition, data collection methods used by WITT’s sources improved over time. For example, START expanded its collection methodology in 2012, resulting in more comprehensive event records. The result is that changes in methodology might have produced unpredictable biases, although the expansion of collection capabilities over time would probably result in a bias toward overstating the growth of terrorist violence over time across the full scope of the data. This bias is unlikely to impact short-term analysis of trends.

Second, due to incomplete public information on disrupted plots, the dataset may not include every plot during the period. However, WITT included every plot that it could verify. Terrorist plotting is studied far less than terrorist attacks, and it is not clear how (or if) any selection bias would impact analysis.

Third, there was often limited information available on motivations for violent incidents. WITT maintained a high standard of proof and excluded incidents for which a political motive could not be determined. Consequently, the dataset may understate the number of attacks. The amount of available information would be expected to increase over time, leading to another bias toward overstating the growth of terrorist violence over time across the full scope of the data. This bias is also unlikely to impact short-term analysis of trends.

Despite these limitations, WITT believes that the dataset of 750 events offers a fair representation of terrorism in the United States from January 1, 1994, to July 4, 2025.

Codebook

YEAR, MONTH, DAY

These columns record the date on which the incident occurred. In the case of plots, WITT recorded the date on which the plot was interrupted, such as the date of arrest.

STATE, CITY, LAT, LONG

These columns record the location of the incident by the name of the state and city and by its geographic coordinates. In the case of plots, if a specific target location was unknown, WITT recorded the location at which the perpetrator was apprehended.

SUMMARY

This column provides a brief description of the incident.

ORIENTATION

WITT categorized each terrorist incident into one of five perpetrator orientations: ethnonationalist, jihadist, violent extreme-left, violent extreme-right, and other. Terms such as extreme-right and extreme-left terrorism do not correspond to mainstream political parties in the United States, such as the Republican and Democratic parties, nor do they correspond to the overwhelming majority of political conservatives and liberals in the United States.

Extreme-right terrorists are motivated by ideas of racial or ethnic supremacy; opposition to government authority, believing it is tyrannical and illegitimate and that it infringes on individual liberties; misogyny, including incels (“involuntary celibates”); hatred based on sexuality or gender identity; belief in the QAnon conspiracy theory; opposition to certain policies, such as abortion; or partisan extremism, where violence is justified against political opponents and parties perceived as advancing left-wing agendas. Some extremists on the violent far-right have supported “accelerationism,” which includes taking actions to promote social upheaval and incite a civil war. While perpetrators of some right-wing attacks may have religious ties, such as some within the Christian Identity movement, these attacks are motivated primarily by concepts of white supremacy and are therefore coded as right-wing. Similarly, though (primarily Christian) religious ideology may influence some perpetrators of abortion-related attacks, these traditionally fall under the definition of right-wing terrorism.

Extreme-left terrorists are motivated by an opposition to capitalism, imperialism, or colonialism; Black nationalism; support for environmental causes or animal rights; support for LGBTQ+ rights; adherence to pro-communist, pro-socialist beliefs or “anti-fascist” rhetoric; opposition to government authority, believing it is a tool of oppression responsible for war and social injustices; support for decentralized political and social systems, such as anarchism; or partisan extremism, where violence is justified against political opponents and parties perceived as advancing right-wing agendas.

Jihadist terrorists are motivated by a violent interpretation of Islam that frames global events as part of a struggle between Muslims and perceived enemies of Islam. They justify violence to defend or expand the Muslim community, oppose Western influence and secular governance, and establish political authority under strict interpretations of sharia. Many jihadists view their struggle as part of a global religious war rather than limited to local grievances.

Ethnonationalist terrorists are motivated by ethnic or nationalist goals, including self-determination. In recent decades, issues driving ethnonationalist terrorism in the United States have included political divisions within the Haitian and Cuban exile communities and Puerto Rican independence. While anti-Semitic motives are classified as right-wing, attacks on Jewish individuals or institutions intended as a response to the Israeli-Palestinian conflict are classified as ethnonationalist.

Incidents in which a political motive was established but did not meet any of these criteria were classified as other.

IDEOLOGYSUBTYPE

This column denotes the primary ideological subcategory of the perpetrator within the five broader perpetrator orientations tracked in the Orientation column.

GROUP

This column provides the name of the group, network, or ideology subset to which the perpetrator belonged, if known and if relevant. Individuals perpetrated most attacks, not groups. Not all variable values indicate formal organizations. In cases where attackers were inspired by a specific group but lacked material ties to that group, the label “(inspired)” was appended to the label. For example, nearly all Salafi-jihadist attacks in the United States since 9/11 were carried out by individuals inspired by foreign terrorist organizations but who did not have communication or material or operational support for those groups.

PERPETRATOR SEX

This column notes the sex of the perpetrator(s) involved. In instances where the perpetrator’s identity remains unknown, sex is tagged as “Unknown.”

PERPETRATOR AGE

This column notes the age of the perpetrator(s) involved. In instances where the perpetrator’s identity remains unknown, age is tagged as “Unknown.”

PLOT

This column distinguishes between attacks (“0”) and disrupted plots (“1”) in the dataset. Incidents were defined as plots if a perpetrator demonstrated plans or intention to commit an act of terrorism that was prevented, most often due to law enforcement intervention or failure during the preparation stages (such as explosives detonated during production). Incidents were defined as attacks if action was taken to carry out an act of terrorism. This includes both attacks that succeeded and those that failed.

Consistent with WITT’s definition of terrorism, as described in the previous section, the authors coded credible threats of violence as attacks rather than plots.

LE

This column denotes whether the perpetrator (or at least one of the perpetrators, if multiple) was a current or former member of a U.S. law enforcement agency. If coded “0,” the perpetrator was identified and had no affiliation with law enforcement. If coded “1,” the perpetrator was identified and had an affiliation with law enforcement. If coded “U,” the perpetrator was not identified to the public or the affiliation could not be confirmed.

Since WITT coded if an incident involved at least one perpetrator affiliated with law enforcement, the full extent of participation by individual members of law enforcement is not illustrated in the dataset. Rather, this column tracks the number of incidents that can be linked to current or former law enforcement personnel. Incidents that involved multiple perpetrators affiliated with law enforcement received the same coding as single-perpetrator incidents with these affiliations.

LESTATUS

This column notes the nature of the perpetrator’s relationship to law enforcement if the value of LE is “1.” Categories tracked are “Active” and “Former.” If LE has a value of “0” or “U,” LEStatus is listed as “n/a.”

MIL

This column denotes whether the perpetrator (or at least one of the perpetrators, if multiple) was a current or former member of the U.S. military. If coded “0,” the perpetrator was identified and had no affiliation with the military. If coded “1,” the perpetrator was identified and had an affiliation with the military. If coded “U,” the perpetrator was not identified to the public or the affiliation could not be confirmed.

Note that this variable only tracks connections between perpetrators and the U.S. military. Perpetrators who were connected to foreign armed services—such as Mohammed Saeed Alshamrani, the Royal Saudi Air Force officer who committed a mass shooting at Naval Air Station Pensacola in December 2019—were coded as “0.”

Since WITT coded if an incident involved at least one perpetrator affiliated with the military, the full extent of participation by individual members of the military is not illustrated in the dataset. Rather, this column tracks the number of incidents that can be linked to current or former military personnel. Incidents that involved multiple perpetrators affiliated with the military received the same coding as single-perpetrator incidents with these affiliations.

MILSTATUS

This column notes the nature of the perpetrator’s relationship to the U.S. military if the value of MIL is “1.” Categories tracked are “Active,” “Reserve,” and “Former.” If MIL has a value of “0” or “U,” MILStatus is listed as “n/a.”

TARGET

This column identifies the primary target of the attack or plot.

Example of values include:

• Abortion facility

• Business

• Demonstrators

• Educational institution

• Government

• Journalists and media

• Law enforcement

• Military

• Private individuals

• Pro-life center

• Religious institution

• Transportation and infrastructure

LOCATION TYPE

This column identifies the type of location at which an incident occurs.

Example of values include:

• Church

• Educational institution

• Event venue

• Government building or political office

• Hotel

• Law enforcement facility

• Military base, installation, or facility

• Mosque or Muslim community center

• Public transportation

• Residence

• Restaurant or bar

• Store or business

• Street or public area

• Synagogue or Jewish community center

Location type and target vary in that the location type notes where an attack occurred and the target notes what or who was the object of the attack. For example, if a politician is attacked at their private residence for their role or work in government, the target is marked “Government” while the location is marked “Residence.”

DEMONSTRATION

This column records whether or not an incident was related to a public demonstration. If coded “0,” the attack or plot was not related to a demonstration. If coded “1,” the attack or plot was related to a demonstration. Terrorist incidents related to public demonstrations were defined to include terrorist attacks committed by demonstrators, attacks targeting demonstrators, and attacks intentionally timed to occur alongside demonstrations, often to obscure the identity or the intent of the perpetrators. Public demonstrations include protests, sit-ins, marches, and other public gatherings intended to advance a social or political cause.

WEAPON

This column records the primary weapon used in attacks or expected to be used in plots.

Example of values include:

• Chemical or biological: ricin, anthrax, and other such substances

• Explosives and incendiaries: bombs, Molotov cocktails, arson, and other weapons that create a blast or fire

• Firearms: automatic, semi-automatic, and non-automatic rifles; shotguns; handguns; unknown and other gun types

• Melee: close-contact weapons that do not involve projectiles, typically involving stabbing or bludgeoning, for example, knives, machetes, axes, and hammers

• Vehicle: cars, trucks, vans, sports utility vehicles, and other automobiles, typically used in ramming attacks

• Other: weapons that do not fall into any other category, such as a bow and arrow

VICTKILLED

This column records the number of victim fatalities caused by a terrorist attack. Perpetrator fatalities are excluded.

DBSOURCE

If applicable, this column notes the original source(s) cited by the databases from which WITT compiled incidents.

WITTSOURCE

This column notes the source(s) through which WITT researchers identified the event for inclusion in the dataset. See the data selection section on page 1 of this methodology for more details on each.

Daniel Byman is the director of the Warfare, Irregular Threats, and Terrorism Program at the Center for Strategic and International Studies (CSIS) in Washington, D.C.

Riley McCabe is an associate fellow for the Warfare, Irregular Threats, and Terrorism Program at CSIS.