Table of Contents

For deterrence to be credible, the United States must have policies, forces, and operational plans to employ nuclear weapons in ways that would most plausibly meet its objectives in the event that deterrence failed and an adversary undertook military actions that could not be stopped by non-nuclear means. Adversary use of nuclear weapons poses the clearest such existential threat.

The United States and Russia have long searched for capabilities and plans to bolster their conventional and, now, cyber and other non-nuclear capabilities to defeat conventional aggression. They have also considered limited ways in which to utilize nuclear weapons to compel the adversary to terminate conflict on tolerable terms. And, to reinforce this compellence and to deter the adversary from escalating further, both superpowers have searched for ways to destroy (or degrade) as much as possible the adversary’s capacity to retaliate.

In the United States, the interest in winning (or at least not losing) potential escalatory war led to the concept and planned practice of counterforce “damage-limitation.”1 Damage-limitation is meant to deter by threatening counterforce strikes and missile and air-defense operations to deny the adversary’s capability to win. This concept is distinguished from the more widely practiced and understood concept of “deterrence by punishment.”

During the Cold War, nuclear strikes were the primary planned means of conducting counterforce operations. In recent years, precision conventional strike systems, perhaps paired with cyber operations and ballistic missile defenses, could complement or supplement nuclear strikes on Russian, Chinese, or North Korean nuclear forces. Non-nuclear damage-limiting attacks would be less destructive to the environment and human populations than nuclear strikes would be. Although the United States hopes that such capabilities will strengthen deterrence, they also could be destabilizing. The prospect of non-nuclear U.S. counterforce strikes could drive Russian or Chinese leaders to increase the number and survivability of their forces in ways that could make escalation—inadvertent or purposeful—more likely. This prospect also could drive the North Korean leadership to launch nuclear weapons early in a conflict for fear that United States and Republic of Korea forces otherwise would soon destroy them.

Prudence requires the pursuit of alternative ways to reduce damage beyond preemptive nuclear strikes.

The quest for damage-limiting capabilities and plans is natural for people whose jobs are to deter war by being able to win it.2 So long as the United States and Russia have had the financial and technical resources to develop and deploy new potentially winning weapons systems, they have been tempted to do so. Arms control agreements have constrained such competition to some extent, but the temptation remains. Yet when American political leaders have learned about these war-fighting or “damage-limitation” strategies, they have found them untenable at best and extremely dangerous at worst, although they often have been unable to change them significantly.3 The probability is too high that Russian and (less so) Chinese nuclear weapons would survive in numbers sufficient to devastate the United States. This conclusion reinforces mutual deterrence of nuclear use and the general value of arms control to constrain (if not lower) the costs and instability of nuclear competition.

Even if it were feasible for the United States to destroy so much of Russia’s nuclear arsenal that Russia would be unable to devastate the United States in response, large-scale U.S. (or Russian) nuclear counterforce attacks themselves could cause fires sufficient to produce climatic effects dubbed “nuclear winter,” along with widespread radioactive fallout.4 The resultant loss of agricultural productivity could severely harm the United States and the rest of the world for years at minimum. For reasons that deserve further investigation, it is doubtful that U.S. nuclear planners (along with Congress, among others) have factored such damage into policy analyses and deliberations. This omission is noteworthy as advances in computing power and climate modeling have made such studies more insightful.5 (Chapter 6 presents a clear argument for the United States and others to conduct new studies on probable climatic effects of various nuclear war scenarios.)

Prudence requires the pursuit of alternative ways to reduce damage beyond preemptive nuclear strikes. If the effects of successful damage-limiting strikes against adversary forces would themselves be catastrophic—to the food chain, to nonbelligerent third countries, to other global common goods, and in terms of humanitarian law—then alternatives would be in everyone’s interest (consider the effects of low- and high-yield nuclear detonations as illustrated in figure 2). Moreover, if use of conventional or cyber weapons to kill hardened targets such as nuclear weapons and command and control systems would be likely to exacerbate crisis instability and escalatory pressures, the planned conduct of preemptive attacks on Russia’s or China’s core strategic deterrents should be reconsidered. This would be consistent with the philosophy expressed in the 2013 Department of Defense report to Congress on the U.S. nuclear employment strategy: “[the] United States seeks to improve strategic stability by demonstrating that it is not our intent to negate Russia’s strategic nuclear deterrent, or to destabilize the strategic military relationship with Russia.”6 U.S. and international security will be best served by applying the same guidance to China.

U.S. planners should explore how variations in numbers, explosive yields, targets, and weather conditions would reduce the immediate and indirect harm to civilians and the environment. Such exploration should specifically consider shifting away from preemptive nuclear attacks on hardened Russian and Chinese strategic nuclear weapons and command and control facilities. (Negotiating reductions in U.S. and Russian silo-based missiles and warheads is another way to pursue damage-limitation, as discussed in chapter 6.) In eschewing or greatly diminishing plans to attack hardened nuclear targets, planners would focus instead on the other targets in current war plans. These legally vetted targets are supposed to represent what adversary leaders hold most dear. By not concentrating on preemptively attacking hardened nuclear force and command and control targets, the United States would reduce pressures on its own leadership and those of Russia and China to escalate from regional nuclear exchanges to all-out nuclear war.

Some will argue that a U.S. nuclear posture designed around lower yields would increase the chance that in a regional conflict the United States would choose to escalate from conventional to nuclear forces. They contend that the lower yield would reduce a president’s inhibitions. Even though this position could help deter adversaries (with the thought that a president could plausibly use these weapons), it also could make adversaries conclude they would fare better by using nuclear weapons before the United States strikes them. This prospect could alarm allies and partners, which could then raise questions about alliance resolve. Russia and China could be tempted to make provocative moves to test (or demonstrate) whether U.S. alliances will crack over fears of U.S. nuclear use and ensuing escalation.

These paradoxes probably are inescapable so long as nuclear weapons exist. However, it bears emphasis that even low-yield nuclear weapons are breathtakingly destructive. For instance, the 5–7 kiloton (kt) “low-yield” warhead recently deployed on twenty Trident D5 SLBMs would be ten times more powerful than estimates of the chemical blast that destroyed the port of Beirut in August 2020. If low-yield weapons were used, the risks of escalation would still be grave even if the upper bounds of destruction in the first phase of nuclear war were significantly reduced.

Incoming presidents often have been relatively unversed in the effects of nuclear detonations, targeting, and doctrine and operational plans.7 A new president and his or her civilian advisers should be briefed early in their term on the list of targets, and the damage calculations used to determine civilian casualties and overall damage, including through environmental effects.

The military officers and civilians who provide this briefing should be required to give the president alternative plans focused on minimizing the level of overall destruction while meeting targeting requirements at lower levels of certainty. These plans should include the use of lower yield, conventional, and other non-nuclear weapons, and should explain how the usage of less-destructive weapons changes the likelihood of destroying targets and may reduce climatic and humanitarian consequences. The fidelity of this briefing should be such that a president and his or her advisers can issue detailed guidance on the employment of nuclear weapons early in an administration, to facilitate a nuclear posture or other strategy review.

U.S. Launch Strategy

The United States maintains an ability to launch its ICBMs quickly after detecting an incoming attack. This LUA (launch under attack) option exists because of the vulnerabilities of NC3 and silo-based ICBMs.

Russia maintains the only nuclear force theoretically capable of a disarming first strike against the United States. Russian ICBMs are maintained on alert, and Russia’s large number of deployed warheads means there are enough weapons to attack U.S. silos, air and naval bases, and other leadership or command and control targets. The LUA option (in addition to second-strike forces) is intended to deter any rational Russian leader from attempting a first strike, knowing that U.S. ICBMs will be rapidly launched to avoid destruction, preventing any potential gains from a surprise attack. The less vulnerable U.S. nuclear-armed submarine and aircraft fleets further bolster deterrence.8

Under LUA, an important objective is to enable a president to make and communicate a retaliatory launch decision before his or her weapons or NC3 networks are destroyed, or before he or she could be incapacitated by an incoming attack. By some estimates, the president would have approximately eight minutes from the time of being notified that an attack is incoming to decide to launch U.S. ICBMs before they could be destroyed.9 Before this decision could be made, the incoming attack would need to be identified, assessed as real (based on two separate ground-based and space-based sensors that rely on different physical principles), and analyzed. Then military leaders must present a series of response options for the president to consider as the eight-minute clock begins.10

The time pressure under LUA causes two risks: that the United States would (1) conduct nuclear retaliation based on false warning, and thereby escalate a nuclear war mistakenly; or (2) risk launching an unintentionally disproportionate and escalatory counterattack because the incoming attack was not accurately assessed.11

The possibility of false warning is not hypothetical. Internal failures in the U.S. command and control systems generated a false warning in 1978, and at least on two other occasions.12 Soviet (later Russian) early warning systems generated false alarms in 1983 and 1995 that too easily could have led to mistaken launches.13 Of course, any retaliation based on false warning would be an unmitigated disaster and could lead to further purposeful adversary escalation.

Opting for LUA also may commit the United States to an inflexible response in a potential war with Russia. In order to make LUA effective, military planners have developed preplanned options that vary in numbers and targets. However, these established options would not be particularly responsive to the specific scenario which unfolds. Since ICBMs are not recallable, any LUA attack plan option involving ICBMs could risk launching a disproportionate, ill-configured response leading to unwanted escalation.

Emerging trends increase these risks.14 Notably, the potential for non-nuclear cyber and/or kinetic attacks against NC3-related assets could reduce the information available to the president. This could exacerbate the risk of either failing to respond to a real attack or responding mistakenly to a false warning. Counter-NC3 attacks could also increase the chance of a disproportionate, escalatory response based on false information. Hence, efforts to strengthen U.S. command, control and communication systems should receive top priority and full resources.

Ultimately, the problem that LUA attempts to solve—a massive incoming Russian attack that could knock out U.S. command and control and/or nuclear forces before they can be launched—is increasingly less likely (and was never likely in the first place).15 During a growing crisis or regional conflict, U.S. intelligence and surveillance capabilities, including its early warning system, should readily detect an adversary’s preparations to conduct a disarming attack. U.S. air and sea-based forces would be dispersed to preserve a still potent deterrent. Technical threats to U.S. space-based intelligence, surveillance, and reconnaissance (ISR) and early warning systems are increasing, but this adds to the imperative to improve the resiliency of these systems, as USSTRATCOM should be doing today.16 LUA is not the solution to this problem.

Fortunately, U.S. posture now relies much less on ICBMs than it did when LUA options were first put in place during the Cold War. Most operational warheads are deployed on SSBNs at sea without the risk of decapitation. The shift in recent decades toward a more survivable U.S. strategic deterrent means that LUA makes less sense. Further reductions of silo-based nuclear weapons (recommended in chapters 4 and 6) would extend this positive trend. Possessing a greater number of survivable nuclear forces allows for safer, more reliable launch options.


Ending any dependence on LUA to deter Russia should be a U.S. priority as part of a general strategy to reduce the probability and potential destructiveness of all-out nuclear war.

Several options would move the United States in this direction.

The most important priority, which must be pursued vigorously, is to strengthen the survivability of command, control, and communication systems.

If U.S. leaders are confident in the survivability of submarine forces and command and control links to them, they could then exercise the option to verify and assess an attack on vulnerable land-based forces before irreversibly releasing U.S. weapons in response. This could mitigate risks of mistaken warning and assessments of incoming attacks. In the meantime, before U.S. command and control systems are upgraded, the concern would remain that a Russian attack could incapacitate U.S. leadership and command and control systems and put into doubt the United States’ capacity to counterattack after withstanding the Russian attack.

To redress this risk, without the hazards of launching ICBMs under attack, former vice chairman of the Joint Chiefs and NORAD commander James Winnefeld has suggested that the United States should develop plans and capabilities to decide under attack (DUA). Unlike the immediate LUA option, with DUA, upon detection of an incoming attack, a president could transmit preplanned orders for U.S. strikes with a time delay on their execution.17 A several-hour delay in executing the strike would allow cancelation or adjustment, but also allow commanders of forces most likely to escape an adversary’s attack—generated bombers and at-sea SSBNs—to carry out orders at the appropriate time. If the detected attack were proven to be false or mistakenly assessed, the president could cancel or adjust the preplanned orders. In the event the detected attack were confirmed as correct, surviving nuclear forces could carry out the preplanned retaliation orders. To anticipate the possibility that the president could be incapacitated or unable to communicate to adjust or cancel earlier authorized launch orders, designated successors would need to be “attached” to the NC3 network. To be sure, under wartime conditions there is no guarantee that the president or a successor would survive or be able to cancel or modify a delayed launch order. be no possibility of doing so regardless of the senior leadership’s survival.

The DUA recommendation complements the recommendation to reduce reliance on silo-based ICBMs, discussed in chapters 4 and 6.

The President’s “Sole Authority”

The president of the United States has sole authority to employ nuclear weapons. He or she can order the use of nuclear weapons without the concurrence of anyone else. This has recently become the subject of debate and legislation in Washington. The issues involved are vitally important and transcend the character or behavior of any one president. Bipartisan analysis, debate, and policymaking should consider the question of whether to adjust nuclear launch authority. Should any president have the sole authority to employ nuclear weapons, under any circumstance? Are there circumstances in which time and prudence would allow for such decisions to be made in consultation with designated members of the cabinet and Congress, and if so, should this be the policy? Would changes in nuclear posture along the lines considered above—especially a DUA policy—facilitate a shift toward a system of shared authorization?

We recommend that the U.S. Congress create a bipartisan commission including former presidential national security advisers, secretaries of defense, and military leaders to analyze and make recommendations on this set of issues.


1 2018 NPR, 23.

2 Charles L. Glaser and Steve Fetter, “Should the United States Reject MAD? Damage Limitation and U.S. Nuclear Strategy Toward China,” International Security 41, no. 1 (2016): 49–98,

3 Fred Kaplan has recently recounted how presidents Kennedy, Johnson, Nixon, Ford, Carter, Reagan, Bush, Clinton, Bush, and Obama, often guided by senior civilian advisers, all concluded that plans for massive nuclear strikes on the Soviet Union/Russia were untenable. See Kaplan, The Bomb. The Defense Department’s 2013 report to Congress on the U.S. nuclear employment strategy reflects the Obama administration’s grappling with how to decrease the role of nuclear weapons in counterforce strategy. See “Report on Nuclear Employment Strategy of the United States.”

4 On nuclear winter, see Alan Robock, Luke Oman, and Georgy Stenchikov, “Nuclear Winter Revisited With a Modern Climate Model and Current Nuclear Arsenals: Still Catastrophic Consequences,” Journal of Geophysical Research 112 (2007), DOI: 10.1029/2006JD008235; and Joshua M. Pierce and David C. Denkenberger, “A National Pragmatic Safety Limit for Nuclear Weapon Quantities,” Safety 4 no. 25 (2018), 5, DOI:10.3390/safety4020025. See also Keir A. Lieber and Daryl G. Press, “The New Era of Counterforce,” International Security 41, no. 4 (2017): 9–49,, especially the “Cumulative Consequences for Counterforce” section (pages 22–32).

5 Joshua Coupe, Charles G. Bardeen, Alan Robock, and Owen B. Toon, “Nuclear Winter Responses to Nuclear War Between the United States and Russia in the Whole Atmosphere Community Climate Model Version 4 and the Goddard Institute for Space Studies Model,” JGR Atmospheres 124, no. 15 (2019), DOI: 10.1029/2019JD030509.

6 See “Report on Nuclear Employment Strategy of the United States.”

7 For example, it is unclear the last time a president participated in a nuclear wargame to fully understand the political, societal, and physical effects of nuclear war. High-ranking civilian advisers continue to explore these issues; however, given the president’s sole authority to authorize nuclear employment, his or her unfamiliarity with nuclear operations is problematic, in our view. See Fred Kaplan, The Bomb, 284–90.

8 Jeffrey Lewis, “Is Launch Under Attack Feasible?,” NTI, August 24, 2017,

9 Jeffrey Lewis, “Our Nuclear Procedures Are Crazier Than Trump,” Foreign Policy, August 5, 2016,

10 “Dual phenomenology” refers to a requirement to source warning of an incoming attack from two separate types of indicators. To this end, the United States deploys space-based infrared sensors that can detect a missile’s launch and land-based early warning radars that can track a missile in its flight. See Office of the Deputy Assistant Secretary of Defense for Nuclear Matters, “Nuclear Matters Handbook 2016,” U.S. Department of Defense, 2016, p. 76.

11 Winnefeld, “A Commonsense Policy for Avoiding a Disastrous Nuclear Decision.”

12 William Perry, “All Souls: Humans Will Err Again,” October 26, 2016,,false%20alarms%20will%20happen%20again.

13 Geoffrey Forden, Pavel Podvig, and Theodore A. Postol, “False Alarm, Nuclear Dangers,” IEEE Spectrum 37, no. 3 (2000): 31–39.

14 One area of growing concern is the role that accelerated disinformation could play in a political leader’s decisionmaking process for nuclear use. Nuclear-armed states should ensure their bureaucratic processes take the attendant risks into account in devising launch policies and procedures. See the recent report by Heather Williams at King’s College of London on this topic:

15 “While we continue to posture our forces to deter large-scale attacks, the 2018 NPR also highlighted the importance of deterring limited nuclear attacks on allies and deployed U.S. forces—something both the Obama and Trump administrations considered more likely than a ‘bolt-out-of-the-blue’ attack.” Bureau of Arms Control, Verification and Compliance, “Strengthening Deterrence and Reducing Nuclear Risks,” 1.

16 Sandra Erwin, “STRATCOM to Design Blueprint for Nuclear Command, Control, and Communications,” SpaceNews, March 29, 2019,

17 Winnefeld, “A Commonsense Policy for Avoiding a Disastrous Nuclear Decision.”