Does Religion Cause Violence?

The claim that only about 7% of wars were religious comes from the index of the Encyclopedia of Wars by Charles Phillips and Alan Axelrod (2005), which catalogs 1,763 wars. Counting that index in 2018, the historian Andrew Holt found 121 entries under “religious wars”, about 7%.

Holt found no statement of how the editors chose; the label appears only in the index. An index entry is yes or no, so a war fought partly over religion counts in full or not at all.

Peace researchers at the University of Bradford made a separate audit for the British Broadcasting Corporation (BBC) in 2003. It rated major wars of the past 3,500 years from 0 to 5 for religious motivation. By Scott Atran’s account (Foreign Policy, 2012), more than 60% showed none and fewer than 7% scored above 3.

Isak Svensson and Desirée Nilsson’s Religion and Armed Conflict (RELAC) data (Journal of Conflict Resolution, 2018) cover 420 pairings of a government and a rebel group. Each pair fought with at least 25 battle deaths in a year between 1975 and 2015. A pair counts as fought over a religious issue when either side stated a religious demand at the outset, even if not its main one.

Such pairs were 3% of those active in 1975 and 55% in 2015. Over the whole period, 75% of them set a government against a group with Islamist claims.

Monica Duffy Toft counted civil wars from 1940 to 2000 (Belfer Center, 2006). She found religious ones rising from 19% of civil wars in the 1940s to 43% in the 1990s. Of the 14 still under way in 2000, 7 were religious. In her count religious civil wars lasted about two years longer than others and were about twice as likely to recur, though neither difference was statistically significant. She called them deadlier (International Security, 2007), but her figures show nonreligious ones killing more civilians a year.

From 1989 to 2003, government-rebel pairs in which either side made explicit religious claims were significantly less likely than others to end in a negotiated settlement (Isak Svensson, 2007).

In sub-Saharan Africa, Matthias Basedau of Hamburg’s German Institute for Global and Area Studies judged that eight of the ten intrastate armed conflicts active in mid-2017 had a religious dimension. At least five were “theological”, fought partly over ideas such as religion’s role in the state.

The RELAC data record what the sides said they wanted. Svensson and Nilsson warn that stated positions often differ from underlying interests: conflicts defined by religious claims “may, or may not, have religious causes.” Basedau adds that there is “no such thing as a purely ‘religious conflict.’” William Cavanaugh (The Myth of Religious Violence, 2009) goes further. He argues that religion has no fixed nature that holds across all times and cultures, and that what counts as religious or secular depends on political power.

In 2023 the EU recorded 120 terrorist attacks. Separatists carried out 70 of them (58%). Jihadists carried out 14, of which five were completed. The completed attacks killed six people and injured 12. Europol, whose EU terrorism report gives these counts, calls jihadist attacks the most lethal.

In 2024 attacks fell to 58. Jihadist attacks, 24, were the largest category, ahead of left-wing and anarchist ones at 21. That year jihadist attacks killed five people and injured 18. Of 449 terrorism arrests that year, 289 were for jihadist offences.

In the RAND-St Andrews chronology of international terrorism, Bruce Hoffman found that none of the 11 identifiable groups active in 1968 was religious. By 1995, 26 of 56 were. That year religious terrorists committed 25% of recorded international incidents but caused 58% of the deaths. In 1996, the latest year with complete figures when he wrote, 13 of 46 groups were religious (RAND Review, 1998–99).

Victor Asal and R. Karl Rethemeyer (Journal of Politics, 2008) studied what predicted how many people a terrorist organisation killed. Its size, ideology, control of territory and ties to other groups did, and religious groups were more likely to kill than others. State sponsorship, the group’s age and its host country did not. In Ido Levy’s study (Terrorism and Political Violence, 2023), jihadist groups kill more per attack than other groups, largely because of their attachment to suicide attacks.

James Piazza (Terrorism and Political Violence, 2009) found that casualty rates vary widely among Islamist groups and follow their organisation and goals more than Islamism itself. He separates “strategic” groups, which behave like secular national-liberation or regime-change movements, from “abstract/universal” groups tied to al-Qaeda’s global network.

The French Fondation pour l’innovation politique (Fondapol) counted 66,872 Islamist terrorist attacks worldwide from 1979 to April 2024, which killed at least 249,941 people. It found that 86.3% of the attacks and 88.9% of the deaths took place in Muslim countries. Drawing on many indicators in 2017, Anthony Cordesman of the Center for Strategic and International Studies wrote that most extremist and terrorist violence occurs in largely Muslim states. This violence, he wrote, “overwhelmingly consists of attacks by Muslim extremists on fellow Muslims.”

In the sectarian conflicts examined here, religion marked the sides alongside ethnicity, power and wealth.

Northern Ireland’s Troubles set Catholics against Protestants, and scholars disagree about how much religion itself mattered. Some call the conflict ethno-national and dismiss religion’s role; others call it religious (Gladys Ganiel and Paul Dixon’s account of the field, 2008). Joseph Ruane writes that Ireland’s Catholic-Protestant conflict rests on “multiple, overlapping differences: religious, ethnic, colonial, political.”

The Sutton Index at Ulster University’s Conflict Archive on the Internet (CAIN) lists 3,532 deaths from July 1969 to the end of 2001. Republican paramilitaries, from the Catholic and nationalist side, were responsible for about 58% of them. Loyalist paramilitaries, from the Protestant and unionist side, were responsible for about 29%, and British security forces for about 10%.

By Michael Spagat’s reckoning in PLOS ONE (2024), the Iraq war killed roughly 300,000 people by violence from 2003 to 2023. That is the whole war’s toll, and sectarian killing is one part of it. His reckoning adds combatants to Iraq Body Count’s documented civilian deaths. He judges it might be a fifth too high or, at a stretch, half the true toll.

The U.S. Defense Department called November 2006 the height of sectarian violence; Iraq Body Count recorded about 29,500 civilian deaths that year and 26,100 in 2007. Fawaz Gerges writes that the invasion “polarized the country along Sunni-Shia lines” by destroying state institutions and establishing a sectarian-based political system (ISIS: A History, 2016).

The Rohingya are a Muslim minority in Myanmar. Médecins Sans Frontières estimated that at least 6,700 Rohingya were killed in the first month of what the authorities called “clearance operations” in 2017. At least 730 were children under five. A United Nations fact-finding mission reported that nearly 725,000 had fled to Bangladesh within a year.

The mission found grounds for senior generals to be investigated and prosecuted for genocide. Its report describes anti-Muslim hate speech spread by extremist Buddhist groups, and calls the Rohingya’s lack of legal status the cornerstone of their oppression. In 2022 the U.S. State Department determined that members of the military had committed genocide.

In India, Ashutosh Varshney and Steven Wilkinson collected every Hindu–Muslim riot reported in the Bombay edition of the Times of India from 1950 to 1995. Eight cities with about 5 to 6% of India’s population had 45% of the deaths. Comparing pairs of cities with similar shares of Muslims, Varshney tied the difference between riots and peace to how much civic life Hindus and Muslims shared. Wilkinson found that state governments decide whether to stop riots by whether doing so will win or lose them votes.

Jeffrey Seul (1999) argues that religion meets the need for identity more fully than other sources of meaning, which may partly explain why group conflicts so often follow religious lines. Frances Stewart and her colleagues found inequalities between religious or ethnic groups, in politics, wealth or status, to be an important catalyst of such conflicts.

The published totals for the crusades to the east, 1095 to 1291, run from 1 million (John Hittell, 1874) to 9 million (John M. Robertson, 1902). Andrew Holt collected them. He notes that modern crusade historians tend not to make or trust such totals, because they doubt anyone can count the deaths over nearly two centuries with any precision.

After the sack of Jerusalem in 1099, Latin chroniclers put 10,000 dead in the Temple area. Later Muslim chroniclers put 70,000 or even 100,000. But Ibn al-Arabi, a Spanish Muslim who had been in the city before it fell, reported 3,000 men and women killed in the al-Aqsa Mosque. The historian Thomas Madden, working from that report and the size of the building, puts the dead there at 3,000 to 5,000.

The death tolls popularly blamed on the Inquisition and the witch trials are far above what historians count. David Plaisted collected polemical totals of 50 million to 150 million in 2006. These totals count everyone the papacy is said to have killed, in wars and crusades as well as by the Inquisition. Historians working from the Spanish Inquisition’s records put its executions in the thousands, most of them before 1540.

Executions in person; people condemned after death or in their absence were burned in effigy instead. Monter, Kamen, Del Col and the 1540–1700 count as Karlikowski’s Ereticopedia review gives them; Borromeo as the Associated Press reported him.

The figure of nine million witches put to death goes back to an extrapolation by Gottfried Christian Voigt, a Quedlinburg lawyer and local historian, in 1784. The American suffragist Matilda Joslyn Gage repeated it in 1893 in Woman, Church and State. Brian Levack puts the executions at about 45,000 from some 90,000 prosecutions. Ronald Hutton puts them at 40,000 to 60,000, probably in the lower half. Most were sentenced by secular courts; the Spanish Inquisition condemned 59 witches to death in its 125,000 trials, Agostino Borromeo said in 2004.

In some places religious leaders and practice worked against violence.

The Community of Sant’Egidio, a Christian lay community in Rome, hosted the talks that ended Mozambique’s civil war. With the Archbishop of Beira and an Italian government envoy, it mediated the 1992 Rome accord. The accord held until RENAMO, the former rebels, took up arms again in 2013; new accords followed in 2014 and 2019.

In Northern Ireland, Father Alec Reid, a Belfast priest, facilitated the secret talks between the nationalist leader John Hume and the republican leader Gerry Adams. Those talks laid the foundations for the 1998 Good Friday Agreement.

Jeremy Ginges, Ian Hansen and Ara Norenzayan (Psychological Science, 2009) tested which side of religion goes with support for suicide attacks. In two surveys of Palestinians and an experiment with Israeli settlers, how often people attended religious services predicted support, while how often they prayed did not. In the experiment, a reminder of prayer dampened support for one attack: Baruch Goldstein’s killing of 29 Muslims at prayer in Hebron in 1994.

A fourth study, a survey across six religions in six countries, found the same split for a willingness to die for one’s beliefs joined to blaming other religions.

A 2016 study by Ginges and colleagues (PNAS) gave 555 Palestinian Muslim adolescents a dilemma about saving five Jewish Israeli or five Palestinian children. Fewer favoured the Palestinian children when answering as they thought Allah would than when answering for themselves.

Additional Protocol I to the Geneva Conventions codified two rules of the law of war in 1977. The rules are that attacks may be aimed only at combatants and military objectives, and that harm to civilians must not be excessive for the military advantage expected. Christian just war theorists, the theologian Francisco de Vitoria among them, had argued for both four centuries earlier.

Both totals count famine or war deaths as well as killings.

For Stalin, Stephen Wheatcroft counts about a million deliberate killings and probably about two million deaths through neglect among prisoners and exiles, and excludes the famine of the early 1930s. Timothy Snyder counts about six million deliberately killed, famine deaths in Soviet Ukraine among them, and about nine million with foreseeable deaths from deportation, hunger and the camps.

The Great Leap figures run from demographers’ estimates of about 30 million to Frank Dikötter’s at least 45 million. Some demographers’ estimates are as low as 17 million. The Cultural Revolution’s figures run from John Fairbank’s one million to Ding Shu’s two to three million. Both estimates are among those Song Yongyi collects, which run from half a million to eight million.

By a news report of its findings, a Russian presidential commission chaired by Alexander Yakovlev found more than 85,000 Orthodox priests shot in 1937 alone.

Snyder ties the Soviet killing to “an ideologically informed vision of modernization” and at times to national or ethnic motives. Tim O’Neill, an atheist who writes History for Atheists, calls two arguments wrong. One is that the Soviet regime killed because it was atheistic. The other is that atheism had nothing to do with the regime’s killing, when it demolished churches as part of its “dedication to the imposition of atheism.”

R.J. Rummel, who compiled the 20th century’s killings by governments, found that they rose with a regime’s power: “The more Power at the center, the more killing.”

The strongest case that religion drives violence rests on the findings above:

• since 1975 a growing share of government-rebel pairs has fought over a religious issue;
• religious civil wars became a larger share of civil wars;
• religious terrorist groups are more likely to kill, and jihadist ones kill more per attack;
• attending religious services goes with support for suicide attacks.

Earlier sections answer three of these points. Svensson and Nilsson warn that a religious claim does not show a religious cause. James Piazza found that an Islamist group’s organisation and goals explain its casualty rate more than Islamism does. And in Ginges, Hansen and Norenzayan’s studies, how often people prayed did not predict support for suicide attacks.

Monica Duffy Toft’s newer data (2021) suggest a rise in religious civil wars. Bormann, Cederman and Vogt replied in 2026 that her classification overstates the rise and that their 2017 result stands. Toft answered them in the same issue of the journal.

Sometimes, but rarely on its own. Counted across history, few wars are listed as religious: the index of one encyclopedia labels 121 of its 1,763 wars religious, about 7%, without saying how it chose. Since 1975, though, conflicts fought at least partly over religious demands have grown from 3% of the government-rebel pairs at war to 55% in 2015. And religious terrorist groups have been deadlier than secular ones.

What the counts do not show is religion acting alone. The Religion and Armed Conflict (RELAC) data record what each side said it wanted, and a religious demand is not proof of a religious cause. Cross-country studies find that more religiously diverse countries have been no more likely to have civil wars once income is taken into account. In those studies, poverty, weak states and exclusion from power predict civil war. How fast religious civil war has been rising is disputed by the researchers who count it.

Download the conclusion

The verdict above as one image, with the two numbers that frame it: the share of government-rebel pairs at war over a religious issue, 3% in 1975, 55% in 2015. It is 1,200 by 630 pixels, the size social networks use for link previews, and anyone may download it and post it.

Download the card (PNG, 1,200 × 630)