For decades the United States and its allies have maintained nuclear forces designed for one overriding objective: to deter Soviet military action, particularly an attack against Western Europe. In addition, the United States has sought to prevent the proliferation of nuclear weapons, but this goal was never considered as important as the first.

Now the crumbling of Soviet power and the war in the Gulf have shattered the assumptions that underlie Western nuclear policies. An entirely new danger has arisen: that a disintegrating Soviet Union might lose control over its immense nuclear arsenal. Among the causes of the war in the Gulf, moreover, was the fear that Saddam Hussein would acquire nuclear weapons capable of hitting distant targets and drawing the Great Powers into a nuclear conflict. New policies controlling nuclear weapons must be adopted to reduce as quickly as possible risks of both kinds.

The crucial first step, we shall argue, would be a very deep and swift cut in US and Soviet nuclear forces, which should be reduced from their present total of approximately 50,000 warheads to something on the order of two thousand. To do so would, at one stroke, force the US and the USSR to adopt far less dangerous nuclear strategies and strengthen the global effort to halt the spread of nuclear weapons.

These changes in nuclear policy would be a major step toward a world in which relations among nations would be based on the rule of law, supported by a system of collective security, with the United Nations and regional organizations able to resolve conflicts and keep the peace.

That the peaceful period between the collapse of the Berlin Wall and the arrival of tanks in Kuwait and Vilnius was so brief should dispel any illusions that such a goal is within easy reach. Nevertheless, at least it is clear that no regime in Moscow will be able to recreate the threat of the cold war years. This brutal century, in our view, has taught some lessons that should be applied as we plan for the next: that war between major industrialized states can no longer be perceived as the continuation of politics by other means; that long-term security depends on cooperation, especially with former enemies, as the success of Western policy toward Germany after 1945 demonstrates; and that a more peaceful and civilized international life will depend on the emergence of institutions with the resources and authority to encourage cooperation and to mediate conflict.

Deciding the role of nuclear weapons in a new system of international security is the easier part of the problem. Whether or not one believes that the numbers and disposition of the US and Soviet nuclear forces were appropriate before 1989, it should be clear they are incompatible with the new political realities. Compared to conventional forces, moreover, nuclear forces are designed to be used in extreme and highly unlikely circumstances; they have little impact on domestic life, and no connection with day-to-day international affairs. Presumably these facts explain why it is a matter of indifference to most people that the world’s inventory of tens of thousands of nuclear warheads remains virtually intact, even though of all the results of the cold war it is the one that poses the gravest danger to survival.

The danger of nuclear war has not vanished along with the Soviet conventional threat. Soviet instability, regional conflicts in the Middle East and elsewhere, and other, unforeseeable, situations present a continuing risk of nuclear destruction that is unacceptable. To reduce this risk, the United States must seize the initiative and exploit the opportunities created by new geopolitical relationships. Our proposals suggest both the immediate strategic objectives called for by today’s situation and workable policies for pursuing them.


America’s nuclear policy originated, to a considerable extent, in order to counter three factors: (1) the large conventional armies that the Soviet Union maintained in the center of Europe after 1945, (2) the belligerent character of Soviet military doctrine, and (3) the Soviets’ geographic advantages in a European war. In the early phases of the cold war, the US policy of building large nuclear forces, tactical as well as strategic, was to a substantial degree intended to offset the imbalance in conventional arms. Under the policy of “flexible response,” tactical nuclear weapons were to be used if NATO’s conventional defense forces were in danger of collapse. The missiles and bombers with nuclear warheads that made up the US strategic forces were also to provide “extended” deterrence, i.e., in addition to their primary mission of deterring a strategic attack on the United States, they were intended to deter an attack by Soviet conventional forces on Western Europe as well. The war plans, targets, weapons, and command organization of US strategic forces long reflected these two requirements.

By the late 1960s the Soviet Union had, in effect, achieved parity with the United States in strategic weapons, and it went on to acquire strategic forces that are, if less flexible and sophisticated, at least as powerful as those of the US. In recent years, the emphasis of the arms race has been on “modernization” of both Soviet and US arsenals by improving their versatility and accuracy. These goals are still being pursued by the US and the USSR. They have not been much affected by the geopolitical earthquake of 1989.


The expensive and dangerous process of modernization could continue in isolation from other relations between the US and the USSR because the military goals that each side has tried to meet in recent decades have been largely set by its opponent’s strategic capacities. While there are significant differences between the two strategic forces, both base their war plans on “worst-case” assessments—i.e., they assume the other side would mount the most powerful and effective attack of which it was capable. They largely fail to ask what political motives could plausibly impel the other side to take so desperate a gamble.

This concentration on the adversary’s nuclear forces has an insidious inner logic. Each side has sought to protect itself by acquiring the means to swiftly disrupt the other’s strategic forces and its command systems, with the result that both strategic forces and command systems are more vulnerable than ever. For example, in addition to targeting Soviet missiles the US is now developing warheads designed to penetrate the deep underground command centers built for Soviet leaders. The Soviet ICBM force has long had similar objectives. No wonder that senior officials in both Moscow and Washington have made it known that they place great emphasis on the ability to launch missiles promptly on being warned of an attack. Both governments have made heavy investments to this end.


The present size and composition of both sides’ strategic forces are not only expensive anachronisms but pose a latent threat that can no longer be justified, even by the criteria that once led many to accept the risk of nuclear war as an inescapable evil. Each of the military commands must be able to ensure that weapons will be launched if, and only if, legitimate orders from civilian authorities have been issued. But they must fulfill this responsibility knowing that the command systems themselves, as well as the bombers and land-based missiles, are vulnerable to missiles that take from fifteen to thirty minutes to strike their targets after being launched. As a consequence, in a nuclear crisis each side’s survival depends on the ability of the command systems on both sides to function without gross error, and, above all, to correctly assess information warning of attack, while being under pressure to do all this in a matter of minutes.

Such requirements become all the more menacing when we realize that modern command organizations are extremely complex combinations of hardware, software, procedures, and human beings that have an astonishing tendency to fail disastrously as the result of what the Yale sociologist Charles Perrow calls “normal accidents.” Chernobyl and a number of tragedies on Soviet submarines, ships, and pipelines show that Soviet organizations are especially prone to normal accidents, and we ourselves are hardly immune to them. The Western press has often asked whether warring factions in the Soviet Union might acquire nuclear weapons, and Soviet officials have implied that they have taken steps to eliminate such a risk, a kind of reassurance that can only heighten such concerns.

From a Western perspective, however, the risk that Soviet nuclear weapons will become embroiled in civil war is overshadowed by the risk that the chaotic breakup of the Soviet government could cause the officials who run the Soviet nuclear command system to take actions that would have catastrophic results, especially if the breakup were to occur during an international crisis that threatens Soviet security. Should such circumstances arise, it is not inconceivable that a mixture of misperceptions, confusion, and panic could somehow lead to an unauthorized or inadvertent launch of Soviet missiles. No doubt there is only a small probability that such a chain of events will occur; but the scale of the ensuing tragedy would be so great that, as with other paths to nuclear catastrophe, every effort must be made to eliminate the risk.

In short, the existing strategic forces serve neither the interests of the United States nor the Soviet Union. The only remaining justification for the enormous size of each state’s strategic forces is the other’s strategic forces. Both sides should be willing, therefore, to move rapidly to far smaller nuclear arsenals. The US must take the initiative. It remains possible that a Soviet regime bent on making trouble for the West could come to power, and should that happen it would be best if the USSR were not so prodigiously armed with nuclear weapons.



The principal objectives of the new US nuclear policy should be, first, a rapid and deep cut of Soviet strategic forces so as to decrease greatly the vulnerability to destruction of the remaining US forces; and second, creating conditions that would give the greatest possible support to the effort to control the spread of nuclear, chemical, and biological warheads, and sophisticated means for their delivery.

After the START reductions have been carried out, the US will still have roughly nine thousand deployed strategic warheads and the Soviets about eight thousand. Each will also have many thousands of strategic warheads that they will be forbidden to deploy but that they need not destroy as well as thousands of tactical nuclear warheads. In addition, China, Britain, and France together can be expected to then have over 1,500 fission and thermonuclear warheads, while the “undeclared” nuclear powers (including Israel and India, and probably Pakistan and South Africa) would have a very much smaller number. This, by any measure, will still be an unacceptably dangerous nuclear armory. If the goals we advocate are to be achieved, negotiations much more ambitious than those now underway are needed. These could not succeed unless the American and Soviet governments first recognized that their separate interests would be best served by abandoning the strategies that have shaped their forces in the past.

This shift would replace today’s strategies for fighting a nuclear war with a strategy of minimum deterrence, a policy that would result in nuclear forces that are models of common sense compared to those now held by both sides. This is so because minimum deterrence would explicitly discard the basic goals that advocates of war fighting strategies have vainly pursued: the ability to retaliate immediately and decisively by destroying a major portion of the enemy’s strategic forces so as to limit damage to oneself and to compel the opponent to refrain from further escalation. Logical as these objectives may be in the abstract, in reality their pursuit has only amplified vulnerability on both sides and fueled the arms race. Furthermore, minimum deterrence would not have the deterrence of conventional attacks as one of its purposes. Indeed, the credibility of this aspect of nuclear strategy was always questionable, and its rationale has now been eliminated by the collapse of Soviet power in Central and Eastern Europe. Finally, minimum deterrence would not seek weapons specifically designed to eliminate the Soviet leadership. For such a minimum strategy, the ability to stop any nuclear combat as quickly as possible—which puts a premium on the survival of leaders—would be more important than any other.

Under a policy of minimum deterrence, the forces and their command systems on both sides would be able to survive to the extent that they could guarantee retaliation even after absorbing a surprise attack. Military commanders would be able to defer recommending a nuclear response until they had taken the time to evaluate the nature and purpose of the attack. Stability in a crisis would be greatly enhanced because the existing pressures to launch nuclear weapons when warned of an incoming attack, or when under attack, would be eliminated.

Minimum deterrence would (1) ensure that the side that mounted a nuclear attack would suffer swift destruction on a scale that would exceed anything that has been wrought in wars in the past and would be unacceptable by any standards to a rational opponent; but (2) it would not threaten the destruction of the opponent’s strategic forces or the very survival of civilization. In contrast, current strategies call for almost simultaneous attack on a vast range of military and industrial targets—some five thousand of them—while “withholding” sufficient strategic forces to gain the upper hand in “peace” negotiations. The ceiling on future forces should be defined by the two main characteristics of minimum deterrence we have just mentioned. As for the floor, it should be high enough to make clandestine cheating unproductive and technical breakthroughs futile, and also sufficient to make it difficult for either side to gain a significant advantage by a crash rearmament program should US–Soviet relations turn sour.

The size and composition of the minimum deterrence forces must be determined by a complex process of analysis and negotiation, in which the hardest bargaining would probably not take place between the US and the USSR but would almost certainly go on within the American body politic, among the US allies, and within the Soviet Union. But it seems evident to us that a total inventory some twenty times smaller than each side has today should be sufficient for minimum deterrence—that is, on the order of one thousand tactical and strategic nuclear warheads apiece. This is not meant to imply that global security will forever require such inventories, for the destructive power of these “small” arsenals would still be tens of thousands times larger than that of the Hiroshima bomb. Further reductions should therefore be sought after experience has established that deep cuts of the size we suggest, along with stringent barriers to proliferation, have been able to create a reliable system of nuclear arms control and disarmament.


The bilateral treaty agreements that American diplomacy should now seek from the Soviet Union fall into two groups: one group should reduce sharply the possibility of a nuclear confrontation between the two countries; the other should reduce the ability and desire of other countries to acquire nuclear weapons.

The composition of a nuclear force that would assure minimum deterrence have been much discussed. We believe there is wide potential agreement on what weapons should be included. As we have said, a ceiling on warheads should be established that is about 5 percent of current levels. A combination of submarine, air, and land-based delivery systems, with constraints on multiple warheads, would render futile a “counterforce” attack designed to eliminate the other side’s nuclear weapons. Tactical nuclear weapons should be eliminated from a region extending from the Atlantic to the Urals and from naval vessels as well. There should be verifiable means of ensuring that undesirable kinds of weapons “modernization” not take place. To that end, antisatellite weapons should be banned and ceilings imposed on missile flight tests to slow down innovation and to prevent the deployment of nuclear weapons specifically designed to attack command centers. The Anti-Ballistic Missile (ABM) Treaty would have to remain in effect in order to give both sides confidence that a sudden deployment of strategic defenses could not neutralize the minimum deterrent.

As for nonproliferation of nuclear weapons, an international agreement would have to include the following measures if it were to be effective: ceasing production of fissile materials for weapons; monitoring of all nuclear weapons production facilities along the lines initiated by the INF and START accords; verification of the destruction of warheads; and a sharp reduction in the number and yield of underground tests, with a commitment to subsequently ban all such tests. All but the last of these measures would be necessary to provide confidence that the US and the USSR were adhering to the reductions they would have independently agreed on. But they all are, in addition, highly important for blocking proliferation.

The proposals we have made here would take decades to negotiate were the pace to be that of past arms control talks. But global security would be greatly improved were such measures put into effect promptly. Both sides should consider taking steps toward arms control that could precede formal agreements. Such steps must be reversible should relations with the Soviets become contentious, and they should be conditioned on reciprocal action by the USSR.1

We have hardly mentioned the other “declared” nuclear powers, Britain, China, and France. Since all three target their nuclear weapons on the Soviet Union, at some point they must participate in nuclear arms control if the reductions we propose are to occur; and they should in any case be concerned to strengthen efforts toward nonproliferation. That they are unlikely to join in at the outset of US–Soviet efforts to reduce nuclear weapons is no excuse for the US or the Soviets to postpone that process. The nuclear superpowers have a long way to go before they could plausibly blame any reluctance to reduce their own forces on the smaller nuclear powers.


Nuclear proliferation has been less rapid than was originally feared, in part because Washington and Moscow have cooperated to discourage other countries from acquiring weapons, even when their bilateral relations were far from amicable. Nevertheless, current trends are disquieting. A number of states are apparently close to attaining an “undeclared” nuclear capacity, or may already have it. Furthermore, the Soviet domestic crisis may encourage highly qualified nuclear experts to sell their services on the world market. This could have serious consequences because sophisticated know-how is more valuable for constructing nuclear weapons than purloined devices or plans. Long-term security therefore demands much stronger efforts to stop proliferation. If this is to occur, the nuclear superpowers must demonstrate that they not only preach nuclear abstinence but are dramatically reducing their own nuclear addiction. The policies we propose would make this change of attitude unmistakable.

Demands for nonproliferation have always been handicapped by the double standard inherent in the post-Hiroshima world. That double standard is reflected in the Non-Proliferation Treaty first signed in 1968. The declared nuclear powers that have signed it—Britain, the US, and USSR—are free to build weapons, whereas other nations that have not signed it (e.g., India and Pakistan) would, if they did so, have to adhere to a stringent code of abstention. This double standard cannot be eradicated, but it could and must become less flagrant. The nuclear powers, starting with the US and the USSR, should therefore not only cut back their nuclear forces, but revise as well their attitudes toward international monitoring of their nuclear weapons programs.

With the INF and START treaties, the US and USSR recognized that verification requires the parties to monitor some of each other’s sensitive activities. The verification of the new bilateral US–Soviet agreements we are advocating would require additional monitoring of testing, production, destruction, and disposal of nuclear weapons. The US and the USSR should offer to have a substantial portion of this monitoring performed by an international agency. Were the nuclear superpowers to set this example, the other three countries that have declared themselves as nuclear powers would come under pressure to follow suit. All permanent members of the UN Security Council would then have committed themselves to international monitoring.

Verifying a more stringent system for controlling nuclear weapons presents difficult technical and political problems. Safeguards against the diversion for military purposes of fissile materials from civilian activities are the responsibility of the International Atomic Energy Agency in Vienna. But the IAEA may only inspect “declared” facilities, and has no authority to investigate other sites even if they are suspect. Nor can the IAEA monitor the military activities of the declared nuclear powers. It is questionable whether the IAEA should be adapted to perform these tasks, for that could undermine its ability to perform its current mission.

A new international institution with its own means for space surveillance may thus be needed for monitoring proliferation of weapons of mass destruction. Responsibility for directing the global nonproliferation program should then be turned over to the UN Security Council.

The Comprehensive Test Ban Treaty (CTBT)—which has not so far been endorsed by any of the declared nuclear powers except the USSR—is often seen as indispensable to nonproliferation of nuclear weapons. This is not so: the Non-Proliferation Treaty will survive even if a Comprehensive Test Ban Treaty is not signed, and a test ban could not stop proliferation by itself since weapons development could still go forward without testing. Nevertheless, by signing the CTBT the nuclear powers would be proclaiming that they have given a high priority to nonproliferation, and we strongly support a comprehensive test ban for this and other reasons. The technical and political issues involved lead us to suggest that the US propose a treaty that would, for a few years, set a limit of at most fifteen kilotons for each test (as compared to the current 150 kiloton limit) with a low annual ceiling (of perhaps five) on the number of tests; the treaty should specify that after the interim period, a comprehensive test ban will come into effect.2

Truly deep cuts in Soviet and American nuclear weapons and comprehensive bilateral agreements along the lines we have described would, for the first time, fulfill the commitment of the nuclear superpowers to “the cessation of the nuclear arms race and to undertake effective measures in the direction of nuclear disarmament” which they accepted when the nonproliferation treaty went into force in 1970.

These steps by the US and USSR, together with nuclear arms reductions by the other nuclear powers, and a strengthened system to stop proliferation, would greatly increase the obstacles to “going nuclear.” They would provide a new basis for both domestic and international opposition to the spread of nuclear weapons, and also complement current efforts to constrain the spread of chemical and biological weapons and of ballistic missiles. But it would be naive to presume that all nonnuclear states would thereafter refrain from seeking nuclear weapons if the nuclear powers—which include most of the world’s strongest states—were to claim that their own security requires “small” nuclear arsenals as large as those we have described here. The barrier to proliferation of weapons of mass destruction can be maintained for the long term only if the international community creates increasingly credible and enforceable guarantees against aggression for all states, accompanied by further substantial cuts in the number of warheads held by the nuclear powers.


America’s existing strategic forces were planned to deal with threats that differ fundamentally from today’s two primary sources of nuclear danger—an unstable Soviet Union armed to the teeth with nuclear weapons, and a world moving toward greater proliferation of such weapons. The United States must therefore put forward new policies that address the dangers that now exist, rather than those of yesterday.

If a shift toward a policy along the lines we have proposed took place, enormous benefits would follow, for the administration, for the Western alliance, and the world at large—benefits in resources saved, and above all, in true security for the living and the unborn. No government other than that of the United States is in a position to take the first steps in this direction, and no one but the President himself can effectively provide the leadership required. Continuing with outdated policies now would come to be seen as a deep failure on the part of the US even if the dangers that exist do not lead to tragedy. For the first time in four decades the opportunity has arisen to dramatically reduce the risk of nuclear war. It must be seized.

This Issue

June 27, 1991