Table of Contents

On July 15, 1948, three weeks after the Soviet Union had begun its blockade of Berlin, the United States announced the dispatch of B-29 bombers to Great Britain and Germany on what was officially described as a training exercise but was actually a message to Moscow. Although the intended meaning—that the United States was prepared to use nuclear weapons to defend Europe—would likely have been obvious to the Soviet leadership, the administration of U.S. president Harry Truman took pains to drive its point home. Official press releases described the aircraft as “atomic-capable.”1 At the same time, a New York Times article based on “authoritative sources” reminded readers that it was B-29 bombers that had dropped nuclear weapons on Hiroshima and Nagasaki.2 While U.S. officials did not lie, the overall impression they created was misleading. Although all B-29s were theoretically capable of carrying an atomic bomb, the specific aircraft sent to Europe had not been modified to do so. In other words, the United States conveyed its first nuclear threat with weapons that were incapable of enacting it and, in so doing, created a risk that Moscow could have interpreted this threat as a bluff.

Fourteen years later, in the summer of 1962, in the run-up to what would turn out to be the Cold War’s most dangerous moment, the Soviet Union shipped about eighty coastal defense cruise missiles and their nuclear warheads to Cuba, along with the medium- and intermediate-range ballistic missiles (IRBMs) that sparked the Cuban Missile Crisis. U.S. intelligence spotted the cruise missiles but incorrectly assessed them to be conventionally armed.3 As a result, plans for the invasion of Cuba, which the U.S. military drew up during the crisis, were based on intelligence that seriously underestimated Soviet capabilities to defend the island.

In 2017, as U.S.–North Korean tensions spiraled more than twenty-five years after the Cold War’s conclusion, the United States conducted a series of exercises around the Korean Peninsula involving B1-B bombers. These operations were explicitly intended to be a “demonstration of U.S. resolve,” but the signal was meant to be a purely conventional one: B1-B bombers ceased to have a nuclear mission in 1994 and were subsequently modified so they could not carry nuclear weapons.4 Yet, if its statements can be believed, North Korea misinterpreted the signal. Pyongyang claimed that the B1-B was nuclear-capable (if not actually nuclear-armed at the time), describing one exercise as a “nuclear bomb dropping drill” and another as a “surprise nuclear strike drill,” while branding the U.S. military as “nuclear war maniacs.”5

In a crisis or conflict, pre-launch ambiguity could create serious risks of escalation by leading one state to misjudge its opponent’s willingness to use nuclear weapons or that opponent’s nuclear or nonnuclear capabilities.

In each of these three events, which span the nuclear age, ambiguity about whether delivery systems were nuclear-armed generated a significant risk of escalation prior to any use of those weapons. Indeed, that risk could recur in a future U.S.–North Korean crisis. Yet almost all of the academic and policy debates surrounding the consequences of such warhead ambiguity have focused on risks after an attack has been initiated (but before the warhead has detonated)—that is, on post-launch ambiguity. Controversy has surrounded, for example, the United States’ program to build a new nuclear-armed cruise missile, the Long-Range Standoff weapon, because of fears that Russia might wrongly conclude that a conventional cruise missile fired in anger was nuclear-armed and quickly respond in kind.6 Fortunately, post-launch ambiguity has remained only a theoretical possibility (with the probable exception of the U.S. nuclear attack on Hiroshima in 1945 when a nuclear-armed aircraft was presumably misperceived as being nonnuclear).7 By contrast, pre-launch ambiguity has been relatively prevalent, and its associated risks therefore deserve systematic analysis.

In a crisis or conflict, pre-launch ambiguity could create serious risks of escalation by leading one state to misjudge its opponent’s willingness to use nuclear weapons or that opponent’s nuclear or nonnuclear capabilities. Nuclear use could be a direct result; or in less extreme circumstances, an escalating series of moves and countermoves—threats, signals, and conventional military operations—could plausibly result in nuclear use. While estimating the likelihood of such escalation is extremely difficult (if not impossible), the potential dangers are so extraordinary that the risks demand attention. After all, a conflict that turned nuclear could cause catastrophic societal destruction, including deaths numbering in the tens, or even hundreds, of millions.

For the sake of concreteness, it helps to analyze the dangers in the context of a crisis or conflict between the United States and China or the United States and Russia. The risks associated with pre-launch ambiguity in such a scenario are particularly significant and also intensifying for three reasons.

First, growing geopolitical tensions are raising the likelihood of the kinds of serious crises or conflicts in which pre-launch warhead ambiguity could contribute to rapid escalation. Reflecting these tensions, the U.S. Department of Defense has identified its “principal priorities” to be “long-term strategic competitions with China and Russia.”8 Beijing and Moscow, meanwhile, put the United States at the center of their own defense planning.

Second, improvements in technology are exacerbating the risks, including by enabling China and Russia to develop longer-range ballistic and hypersonic boost-glide missiles that can carry nuclear or nonnuclear warheads. The reach of these missiles exacerbates the potential threat they could pose to the U.S. territory, forces, and assets. In a crisis or conflict, concerns that they might be used could increase pressure on Washington to act quickly by, for example, preemptively attacking missiles that it believed were conventionally armed. If, as a result of ambiguity, the United States’ actions were based on incorrect information, they could have unforeseen and dangerous consequences, such as the unintentional destruction of nuclear-armed missiles.

Third, particularly in the case of the United States and Russia, growing tensions are undermining cooperative measures that help to prevent crises and conflicts and mitigate the risk of escalation should a war occur. The United States and the Soviet Union, later Russia, led efforts to create the Euro-Atlantic security architecture, mostly in the years around the end of the Cold War. This system of agreements is now in a state of accelerating decay because of selective implementation, outright noncompliance, and abrogation. Agreements designed to make a war less likely by limiting military forces and enhancing transparency look ever more inadequate. The 1990 Conventional Forces in Europe Treaty is entirely defunct. The 1992 Open Skies Treaty and 2011 Vienna Document are less than completely functional. (Indeed, the United States has informed allies that it will withdraw from the Open Skies Treaty unless they find a way to assuage U.S. concerns about its effectiveness.)8 The demise of the 1987 Intermediate-Range Nuclear Forces (INF) Treaty, following Russian noncompliance and U.S. withdrawal, has a more direct effect on the problem at hand: The end of the prohibition against deploying ground-launched cruise and ballistic missiles with ranges between 500 and 5,500 kilometers (310 and 3,400 miles), irrespective of how they are armed, has removed a key legal barrier to the development—by Russia, in particular—of longer-range ambiguous missiles. Thus, the risks associated with pre-launch ambiguity have become even more acute.

Key Concepts

Before assessing the risks, it is useful to establish some basic concepts and definitions. Warhead ambiguity complicates the task of characterizing an adversary’s weapons—that is, determining how they are armed. Of course, even when ambiguity arises, a state may characterize the weapons correctly. However, if the state does not, uncertainty or mischaracterization can lead to a false positive (misidentifying a nonnuclear weapon as nuclear) or a false negative (misidentifying a nuclear weapon as nonnuclear). For example, per the events described earlier, if Soviet leaders had concluded that the B-29 bombers deployed to Europe in 1948 were nuclear-armed, their belief would represent a false positive. Conversely, the United States’ misperception that some Soviet cruise missiles deployed to Cuba during the 1962 crisis were conventional represents a false negative.

The case for worrying about pre-launch warhead ambiguity is presented in chapter 2. Launch is defined here as the moment, after an employment decision is made, when a delivery system is set in motion toward its target. Thus, in the case of a mobile missile, launch generally occurs when the missile separates from its launcher, not when the launcher is dispersed.9 For an aircraft, launch occurs at takeoff if the decision to attack a target has already been made, but it can also occur later if that is when an employment decision is made.

The likelihood of mischaracterization or uncertainty is greatest for ambiguous delivery systems, which are described in detail in chapter 3. Such delivery systems include dual-use weapons (for which both nuclear and conventional warheads are available and which can carry either) and superficially similar nuclear and nonnuclear weapons.10 Both categories of weapons contribute to the growing entanglement of the nuclear and nonnuclear domains.11 The risks of pre-launch ambiguity are greatest with ground-launched mobile missiles and aircraft because they can be visibly deployed for prolonged periods—hours, days, or, in the case of missiles, even weeks.

Ambiguity can induce uncertainty or mischaracterization at two separate bureaucratic levels, as discussed in chapter 4. Intelligence analysts may misinterpret—or be unable to conclusively interpret—the available information about how ambiguous delivery systems are armed. Especially in a crisis or conflict, such information could be limited in both quality and quantity. If analysts cannot reach a definitive conclusion, decisionmakers may feel that prudence requires them to assume that ambiguous weapons are nuclear-armed.

In turn, mischaracterization could spark escalation in two ways, as outlined in chapter 5. Most importantly, it could lead a state to underestimate or overestimate its opponent’s willingness to use nuclear weapons, which could frustrate bargaining or lead to an overreaction. Alternatively, a state could develop an inaccurate picture of its adversary’s nuclear or nonnuclear military capabilities and launch an operation whose dangers it had underestimated. Uncertainty, by contrast, would be less escalatory than mischaracterization because it would not necessarily lead to misperception. But uncertainty is also less likely to occur than mischaracterization because decisionmakers tend to assume, without clear evidence to the contrary, that ambiguous weapons are nuclear-armed.

The focus here is on inadvertent escalation—that is, escalation resulting from unintended mischaracterization or uncertainty. Because Beijing, Moscow, and Washington command diverse and mature nuclear arsenals (even if Beijing’s nuclear force is small), they have little incentive to bluff, as the United States did in 1948.12 That said, as discussed briefly in chapter 5, China, in particular, may have a different reason to exploit warhead ambiguity: to enhance deterrence, it may be trying to increase the risk that strikes against its conventional forces would inadvertently destroy some of its nuclear weapons and spark escalation. (Outside of the U.S.-Russia-China triangle, intended ambiguity may be a more prominent feature of states’ strategies; Pakistan, for example, may seek to give the impression that at least one of its conventional missiles is dual-use.)13

Just as a state might accompany a nuclear signal with a warning designed to clarify its intent, it could issue reassurances about operations involving conventionally armed ambiguous weapon systems to reduce the likelihood that an adversary might mischaracterize the weapons as nuclear-armed.

Recognizing the escalation dangers created by pre-launch warhead ambiguity, China, Russia, and the United States ought to assess whether the benefits of ambiguous delivery systems outweigh the risks. However, as described in chapter 6, they are extremely unlikely to forsake such weapons, particularly for cost reasons. Realistically, though, states could assess the risks and develop practices to mitigate them, either unilaterally or, better still, cooperatively. For example, just as a state might accompany a nuclear signal with a warning designed to clarify its intent, it could issue reassurances about operations involving conventionally armed ambiguous weapon systems to reduce the likelihood that an adversary might mischaracterize the weapons as nuclear-armed.

Many of the issues discussed below raise questions of interpretation and perception. In assessing an adversary’s capabilities, how do intelligence analysts interpret ambiguous, incomplete information? If analysts cannot draw firm conclusions, how do decisionmakers deal with uncertainty in planning and conflict management? In peacetime, do decisionmakers regard escalation risks as sufficiently serious that they are motivated to put in place risk-reduction measures? To address such questions, this report draws on experimental psychology and, in particular, on a growing understanding of the ways in which biases and heuristics affect perception and decisionmaking under conditions of uncertainty.14 While the strength of such tendencies differ from person to person, this approach is nonetheless helpful in better understanding how typical analysts and decisionmakers might behave.


1 Gregg Herken, The Winning Weapon: The Atomic Bomb in the Cold War, 1945–1950 (New York: Alfred A. Knopf, 1980), 259.

2 “60 B-29’s Ordered to Fly to Britain,” New York Times, July 16, 1948, 4.

3 David G. Coleman, “The Missiles of November, December, January, February . . . The Problem of Acceptable Risk in the Cuban Missile Crisis Settlement,” Journal of Cold War Studies 9, no. 3 (Summer 2007): 18–19.

4 Dana W. White, “U.S. Flies B1-B Bomber Mission off of North Korean Coast,” Press Release, U.S. Department of Defense, September 23, 2017,

5 Korean Central News Agency (KCNA), “U.S. Imperialists’ Reckless Military Provocation Under Fire,” May 30, 2017,; KCNA, “U.S. Reckless Nuclear War Frenzy Slammed,” August 9, 2017,; and KCNA, “U.S. Surprise Nuclear Strike Drill in S. Korea,” November 3, 2017,

6 See, for example, Matthew Costlow, “The New Nuclear Cruise Missile and the Stability Argument,” RealClear Defense, February 8, 2016,; and William J. Perry and Andy Weber, “Mr. President, Kill the New Cruise Missile,” Washington Post, October 15, 2015, Concerns about ambiguity emerged before plans to develop a conventional variant of this missile (which has an uncertain future).

7 What is probably the most relevant example of post-launch ambiguity involving a nonnuclear system that was misidentified as nuclear-armed occurred in 1995 when Russia briefly concluded that a Norwegian sounding rocket was a nuclear-armed Trident D-5 sea-launched ballistic missile. However, given this incident’s occurrence in peacetime, the lack of direct and detailed evidence about how Russia’s military and civilian leaders perceived the incident, and subsequent improvements to the Russian early-warning system, there has been an unsurprising lack of consensus about its significance. Theodore A. Postol, “An Evaluation of the Capabilities and Limitations of Non-Nuclear-Armed Trident Ballistic Missiles for Short-Time Conventional Strikes,” presentation, Washington, DC, October 5-6, 2006, 10; and Keith Payne, Thomas Scheber, Mark Schneider, David Trachtenberg, and Kurt Guthe, Conventional Prompt Global Strike: A Fresh Perspective (Fairfax, VA: National Institute Press, June 2012), 48,

8 U.S. Department of Defense, Summary of the 2018 National Defense Strategy of the United States of America: Sharpening the American Military’s Competitive Edge, [2018], 4,

8 Joe Gould and Aaron Mehta, “US to Europe: Fix Open Skies Treaty or We Quit,” Defense News, November 21, 2019,

9 Launch could occur after separation in the case of a cruise missile with loitering capability.

10 Here, the term “dual-use” is used instead of “dual-capable” because many weapons systems could carry either nuclear or nonnuclear warheads, but only a subset are actually assigned both nuclear and nonnuclear missions.

11 For more on entanglement, see James M. Acton, “Escalation Through Entanglement: How the Vulnerability of Command-and-Control Systems Raises the Risks of an Inadvertent Nuclear War,” International Security 43, no. 1 (Summer 2018): 56–99; and James M. Acton, ed., Entanglement: Russian and Chinese Perspectives on Non-Nuclear Weapons and Nuclear Risks (Washington, DC: Carnegie Endowment for International Peace, 2017),

12 Official U.S. sources have occasionally stated or implied that the B1-B bomber is still dual-use, but such instances are almost certainly mistakes and not part of a deliberate policy of ambiguity. For an example, see Jeffrey Lewis, Twitter post, December 20, 2018, 5:29pm,

13 Feroz Hassan Khan, Eating Grass: The Making of the Pakistani Bomb (Stanford, CA: Stanford University Press, 2012), 237–238.

14 For classic overviews, see Robert Jervis, Perception and Misperception in International Politics, new edition (Princeton, NJ: Princeton University Press, 2017); and Daniel Kahneman, Thinking Fast and Slow (New York: Farrar, Straus and Giroux, 2011).