In “Conventional War Between Nuclear Powers”, published in the Journal of Strategic Studies on 29 October 2026, Carter Malkasian and Zachary Constantino examine what the history of armed clashes can tell us about the likelihood and character of war between nuclear-armed states. They challenge the assumption that nuclear deterrence has prevented such states from fighting one another. Since 1945, nuclear powers have repeatedly engaged in direct military action, marked by “a low level of violence that sometimes ramped up to higher intensity”. The authors examine how leaders have tried to contain such clashes and how accidents or miscalculations have threatened escalation.
The authors begin by contrasting the large-scale conventional wars anticipated by defence establishments with the expectations of scholars who emphasise nuclear deterrence. US military planning has long prepared for major conflicts, from a possible Soviet invasion of Western Europe to a high-intensity air and naval war with China over Taiwan. Scholars such as Robert Jervis, however, argue that “the fear of nuclear war does deter the other side from much more than nuclear attack”. Kenneth Waltz similarly maintains that, when vital interests are at stake, the prospect of retaliation means “victory in war is too dangerous to fight for”. Political leaders were divided. Some, including Josef Stalin, Mao Zedong, John F. Kennedy, and Robert McNamara, were “reluctant to trust that nuclear weapons on their own could prevent a major war” and built up conventional forces.
They define an “armed clash” as “one or more violent engagements by one state against another”. Such episodes may last a day or extend over several years, and can occur during ongoing competition, a crisis, or a war. Drawing on scholarship, published primary documents, Arabic-language sources, and conflict datasets, the authors compile their record, excluding accidental collisions, non-state attacks, espionage, and advisers not directly involved in combat. They acknowledge an incomplete record and difficulties in counting prolonged exchanges. Their qualitative analysis examines the duration and intensity of engagements, their objectives, the constraints on military operations, and any escalation that followed.
The Cold War cases show that nuclear powers fought directly while taking steps to prevent a wider confrontation. The biggest clash between the United States and the Soviet Union occurred during the Korean War, when Soviet MiG-15 pilots fought US and allied aircraft between November 1950 and July 1953. Stalin had initially resisted deploying Soviet pilots because he wanted to avoid a world war with the United States. Their participation was concealed: pilots wore Chinese uniforms, aircraft carried Chinese markings, and operations were restricted largely to “MiG Alley” in northern North Korea. Stalin prohibited offensive strikes against US forces and South Korea. During the Cuban Missile Crisis in 1962, the US blockade and reconnaissance flights brought American forces into contact with Soviet air defences, which shot down a U-2, and with nuclear-armed submarines. Ambiguous instructions to Soviet submarine commanders heightened the risk of accidental escalation.
China’s intervention in the Vietnam War provides another example. Between 1965 and 1970, Beijing deployed around 320,000 personnel to North Vietnam. Chinese anti-aircraft units repeatedly engaged US aircraft, and American bombing killed or wounded approximately 5,000 Chinese troops. Beijing sought to protect its border and prevent North Vietnam from falling to the United States without entering a direct war with Washington. During the 1969 Sino-Soviet border clashes, Soviet leaders considered pre-emptive strikes against Chinese nuclear facilities, while Beijing prepared for a possible attack. Tensions eased once the two sides began talks in October. The Soviet Union also fought Israeli forces during the War of Attrition and the 1973 Yom Kippur War. Soviet personnel in Egypt operated air defences, provided artillery support, and engaged Israeli aircraft. Moscow concealed much of its involvement and restricted operations geographically. The larger Arab offensive in October 1973 nevertheless prompted a serious crisis between Washington and Moscow. The authors ask why the Soviets would “so strenuously avoid confrontation with the United States yet engage in direct combat with Israel”, despite evidence of Israel’s nuclear capability.
South Asia became the principal setting for armed clashes between nuclear powers towards the end of the Cold War and in the decades that followed. India tested a nuclear device in 1974, while Pakistan may have assembled its first bombs during the 1980s. In 1984, India occupied the disputed Siachen glacier. Pakistani military action followed, and skirmishes continued along the glacier and the Line of Control (LoC) for years. The authors estimate as many as 1,240 casualties. In the winter of 1998–99, Pakistani soldiers posing as local rebels occupied positions across a 150-kilometre front in Kargil. India responded with ground operations, artillery, and air strikes, but Prime Minister Atal Bihari Vajpayee prohibited aircraft from crossing the LoC because of the risk of Pakistani nuclear retaliation. Pakistan kept its aircraft on its side. More than 1,700 Indian air sorties followed, and the war caused over 3,000 casualties. Both countries considered further escalation before Indian resolve and the intervention of US President Bill Clinton compelled Pakistan to withdraw. The authors regard Kargil as “a foremost case of combat between nuclear powers below a nuclear threshold”. They also note that it strengthened Indian thinking about limited conventional war with Pakistan, contributing to the development of the “Cold Start” doctrine.
After the Cold War, South Asia remained the main setting for clashes, although US and Russian forces also fought in Syria. In February 2018, Russian paramilitaries and Syrian forces attacked an American position near a gas facility. Despite an established military deconfliction line, Russian commanders denied involvement. American aircraft killed 200–300 Russian and Syrian personnel, after which Moscow requested a ceasefire rather than retaliating. The authors also discuss US-Pakistani exchanges along the Afghan border, the 2020 Galwan clashes that killed 25 Indian and Chinese soldiers, and continued India-Pakistan exchanges. The 2016 Indian “surgical strikes” and the 2019 Balakot crisis brought New Delhi and Islamabad into further confrontation. After Pulwama, India struck a suspected militant facility inside Pakistan. A Pakistani retaliation led to a dogfight and the capture of an Indian pilot, whose return helped prevent further escalation.
What the authors call “the most intense round of combat since Kargil” occurred between India and Pakistan from 7 to 10 May 2025, following an attack in India-administered Kashmir that killed 26 civilians. India struck militant sites in Pakistan-occupied Kashmir and targets inside Pakistan; Pakistan retaliated against Indian military installations. The exchanges involved aircraft, missiles, artillery, and drones. Both sides sought to limit the fighting geographically. An Indian strike on Nur Khan Air Base, a few miles from the headquarters of Pakistan’s Strategic Plans Division, brought the conflict close to a key nuclear command-and-control institution. Pakistan convened its National Command Authority, “possibly to signal that discussions about nuclear retaliation were underway”. A ceasefire followed on 10 May. The authors estimate around 80 deaths, while noting that details remain disputed.
The authors distinguish these clashes from wars involving a nuclear-armed state and a non-nuclear opponent. Their main record therefore excludes the 1973 Arab-Israeli War, the Israel-Iran conflicts of 2025 and 2026, and Ukraine’s attacks on Russian territory. These cases show that large-scale conventional fighting can occur despite nuclear weapons, which the authors find “disquieting for confidence that nuclear powers would never do the same”. Egypt and Syria attacked Israel in 1973; Iran later fired missiles and drones at Israeli targets; and Ukraine struck Russian airfields, oil facilities, and strategic bombers. Russia issued nuclear threats but did not use nuclear weapons. The authors regard these cases as more troubling, but outside the scope of their main comparison.
Drawing the cases together, Malkasian and Constantino identify four patterns. First, armed clashes between nuclear powers have been relatively frequent. They identify 15 clashes since 1945, with fighting occurring in 61 of the 81 years examined. Around five clashes took place between 2015 and 2025, three involving India and Pakistan. Whether this increase will continue or extend beyond South Asia remains uncertain. Second, violence has generally remained limited. Infantry skirmishes, air-defence engagements, artillery exchanges, special operations, and air-to-air combat have been more common than large-scale offensives. Some involved hundreds of aircraft or thousands of casualties, but none became a major conventional war. Third, leaders have tried to contain fighting by limiting objectives, restricting operations geographically, and concealing their involvement. Fourth, such precautions have not removed the risk of escalation. Accidents or transgressions occurred in at least half the clashes, and some leaders undertook risky military actions without adequately accounting for the nuclear consequences. In the authors’ words, “any violent engagement was a step in brinkmanship”.
If past trends continue, they anticipate a few clashes per decade, most likely involving skirmishes, covert operations, missile or drone strikes, and air-to-air combat. Some could become small-scale wars. A large-scale conventional war “would appear quite unlikely, and would be unprecedented, but could not be ruled out”. A US-China war, for example, may not resemble the high-intensity air and naval campaign often anticipated in American planning. Both sides would have incentives to limit operations and preserve opportunities for negotiation. Political and nuclear calculations could matter more than battlefield outcomes.
The policy discussion is directed primarily at the United States. The authors urge US planners to recognise the risk of nuclear escalation, limit objectives, confine operations geographically, avoid inflicting heavy casualties, and preserve room for negotiation. Yet forces designed only for low-intensity clashes would be inadequate for deterrence and conflicts involving non-nuclear adversaries. Preparing at any cost for an improbable all-out war could also divert resources. The challenge is to maintain credible deterrence without allowing worst-case scenarios to dictate force design.
The article’s historical account shows that nuclear weapons have not ended armed conflict between states that possess them. They have influenced the way such conflicts are conducted, the objectives pursued, and the limits leaders place on military operations. This distinction is especially relevant to India. Kargil, Balakot, the clashes with China, and the May 2025 confrontation show that nuclear-armed adversaries can engage in direct military action without immediately moving towards an all-out war. However, restraint should not be mistaken for a dependable mechanism of escalation control. The strike near the headquarters of Pakistan’s Strategic Plans Division in May 2025 illustrates how conventional operations can approach sensitive nuclear infrastructure, raising risks that extend beyond the immediate battlefield. By placing South Asian cases alongside engagements involving the United States, Russia, China, and Israel, the article also helps assess whether the patterns observed in the subcontinent reflect a wider feature of relations between nuclear powers. Its argument is not that major war is impossible, but that the more common danger may lie in recurring limited clashes whose escalation is difficult to predict. The authors acknowledge that “history cannot predict the future and is just one viewpoint”.
