In the wake of the attacks on the World Trade Center and the Pentagon, America has been mourning its dead and tending to its wounded. But the country also has been building up an angry resolve to respond to this outrage against humanity, and a pragmatic resolve to reduce its vulnerability to future attacks. The world has seen just how terrible the consequences can be when terrorists have the hatred to murder innocent civilians, the resources to coordinate and conduct systematic operations, and the fanaticism to sacrifice their own lives. The evil genius who conceived of using a passenger airplane in kamikaze mode calculated that its 200,000 pounds of jet fuel would make it a weapon of mass destruction. And so it was, with more than 3,000 deaths resulting from each plane used against the World Trade Center, more than ten times the fatality rate caused by past attacks with truck bombs.

The United States can take many actions to make this sort of attack more difficult to carry out, and it will do so, despite the inconvenience and expense. But as Washington moves to reduce the vulnerabilities exposed by the last strike, it should also try to anticipate the next one. As deadly as the World Trade Center disaster was, it could have produced a hundredfold more victims if the terrorists had possessed nuclear or biological weapons. And the future threat could come from hostile nations as well as terrorists.

Nuclear or biological weapons in the hands of terrorists or rogue states constitute the greatest single danger to American security—indeed, to world security—and a threat that is becoming increasingly less remote. Several nations hostile to the United States are already engaged in covert programs to develop nuclear weapons, and multinational terrorist groups have demonstrated both by word and by deed that their goal is to kill Americans and destroy symbols of American power. Such terrorists have escalated their methods from truck bombs to the near equivalent of a tactical nuclear weapon, and they clearly have the motivation to go further up the ladder of destruction. Indeed, Osama bin Ladin has told his followers that the acquisition of weapons of mass destruction is a "religious duty." The only question is whether they will succeed.

Since the end of the Cold War, the barriers to success have been lowered. The know-how for making nuclear weapons is increasingly available through the Internet. Security controls on the huge supply of nuclear weapons (which number in the tens of thousands) and fissile material (amounting to hundreds of tons) are becoming increasingly uncertain. And the thriving black market in fissile material suggests that demand is high. In the next few years this combination of forces could result in a nuclear incident with results more catastrophic than the destruction wreaked by the Hiroshima and Nagasaki bombs, which together killed an estimated 200,000 people.

A nuclear attack's capacity for destruction is familiar by now, but recent simulations indicate that an attack with smallpox germs could cause just as many deaths. Furthermore, there is good reason to fear that biological weapons could become the weapon of choice for terrorists. They can be produced without the massive infrastructure required for their nuclear counterparts, and biotechnology pharmaceutical developments are proliferating these production techniques. Hostile groups that cannot develop their own weapons, meanwhile, may be able to buy them through illicit channels. The Soviet Union produced a large supply of biological weapons during the Cold War, some of which may still be available. China, North Korea, and Iraq have all had biological weapons programs, as did the Aum Shinrikyo cult in Japan, which in 1995 released a chemical weapon, sarin, in a deadly attack on Tokyo's subways.

Finally, the threat posed by long-range missiles has received much attention. But a long-range missile in the hands of a hostile force does not pose a significant new danger unless the missile has a nuclear or biological warhead. Nuclear and biological weapons, in contrast, are dangerous even in the absence of missiles, since they can be delivered by a range of methods, including trucks, cargo ships, boats, and airplanes. Indeed, given its attractions, covert rather than overt delivery is not only feasible, it is the most likely method of attack.

Considering the level of catastrophe that could occur in a nuclear or biological attack, mitigating such threats should be an overriding security priority today, just as heading off a nuclear attack was an overriding priority during the Cold War. In that era the United States essentially depended on a single strategy: deterrence. Now it can add two other strategies to the mix—prevention (curbing emergent threats before they can spread) and defense. Rather than relying exclusively on any one strategy, the sensible approach is to deploy a balanced mix of all three. Missile defense should be one element of national policy, but if the single-minded pursuit of it conflicts with programs designed to curb proliferation and strengthen deterrence, it could decrease our own security rather than increase it.


Prevention is the first line of defense against the proliferation of weapons of mass destruction, but it requires cooperation from the other nuclear powers. Any actions that the United States takes to stop the spread of weapons can easily be nullified if Russia, for example, decides to sell its nuclear technology, weapons, or fissile material. Russian leaders know that it is in their national interest to fight proliferation. But they may at some point be torn between their security interests and the need to earn hard currency. This financial incentive might delude them into thinking that the sale of commercial nuclear technology to Iran, for instance, would not facilitate Iran's development of nuclear weapons.

The cooperation necessary to prevent proliferation is manifested through treaties already in force, such as the nuclear Nonproliferation Treaty, the Strategic Arms Reduction Treaty (START), and the Biological Weapons Convention; through treaties not yet implemented, such as the Comprehensive Test Ban Treaty, START II, and START III; through bilateral and multilateral agreements, such as the Trilateral Agreement (among the United States, Russia, and Ukraine), the Agreed Framework (between the United States and North Korea), and the missile agreement under discussion with North Korea; and through cooperative programs to reduce nuclear risks and manage Cold War-era nuclear arsenals, such as the Nunn-Lugar program with Russia and other former Soviet states.

Many of these programs have been quite successful. The Nunn-Lugar initiative, for example, in concert with start and the Trilateral Agreement, has already been responsible for the dismantling of more than 5,000 nuclear warheads and the complete elimination of nuclear weapons in Ukraine, Kazakhstan, and Belarus. And as a result of programs designed to immobilize or commercialize leftover Soviet plutonium and weapons-grade uranium, material that was once intended for Soviet bombs will soon meet half the supply needs of American power reactors.

To prevent future proliferation, the United States should sustain and build on these programs, by extending the Nunn-Lugar efforts to tactical nuclear weapons, for example, and by funding proposed efforts to immobilize plutonium. But some weapons materials have already spread, and future prevention efforts will not always be effective (as evidenced by the expelling of U.N. weapons inspectors from Iraq). So the second line of defense must remain deterrence.

Even if START II and START III were fully implemented, the United States would still be left with a nuclear force capable of destroying any nation reckless enough to use nuclear weapons against it. In particular, a nuclear attack using ballistic missiles would be instantly tracked to its place of origin and thus invite immediate retaliation by U.S. nuclear forces—a fact known by all.

Some worry that a nation with nuclear weapons might attack a U.S. ally with conventional weapons, believing that Washington would not honor its defense commitment for fear of provoking a nuclear attack on U.S. cities. But any such move would be a serious mistake, since the United States would respond in kind—with its own conventional military forces—to a conventional attack on an ally. The aggressor might then threaten a nuclear strike but would have to contemplate, once again, the certain knowledge of immediate and catastrophic retaliation. So long as the United States maintains strong conventional forces, therefore, the threat of nuclear extortion reverts to the classic deterrence scenario. Moreover, if threatened, the United States has the capability to destroy a hostile nation's launch sites, storage sites, and production facilities with its long-range, precision-guided, conventionally armed weapons—and others know it. Whatever Washington's stated policy, therefore, no hostile nation could rule out the possibility that the United States would strike back if attacked.

In short, the United States has a powerful and credible deterrent involving both nuclear and conventional weapons, which should make a direct nuclear attack or nuclear extortion by a nation very unlikely. The chance still exists, however, that a hostile nation armed with nuclear or biological weapons could end up under a leader who is mentally unbalanced or who miscalculates the consequences of his or her actions. And a terrorist group is probably less deterrable; its members might believe that an attack could not be traced back to them, or they might even be seeking to die for their cause. Both prevention and deterrence, in other words, could fail in the face of terrorism, and there is always the possibility, however remote, of an accidental or unauthorized launch from another nuclear power. Any of these contingencies would create a catastrophe, so it is reasonable for the United States to seek "catastrophe insurance," much as individuals buy earthquake insurance to cover the possibility that their house might be destroyed by such an event.


The most immediate danger is of a terrorist group delivering a nuclear bomb or biological weapon with a truck, cargo ship, airplane, or boat. Such an attack could be tactically similar to what the United States has already experienced—in the 1993 World Trade Center bombing, the 1996 explosions at the Khobar Towers in Saudi Arabia, the attacks on the U.S. embassies in Kenya and Tanzania in 1998, and last fall's bombing of the U.S.S. Cole—and the ultimately responsible parties would be equally difficult to identify. The probable culprit would be a well-organized multinational group, acting with direct or indirect support from one or more hostile nations. Regular military defense tactics by the United States would be largely irrelevant, since the attackers would conceal the place and time of the strike, and Washington cannot maintain terrorist alerts continuously for the entire nation. The first line of defense against this threat, accordingly, is to develop an intelligence network able to give the government advance warning of an attack so that it can be stopped before it is launched.

As Washington tries to step up its intelligence activities, however, it will face two barriers: the restrictions imposed on U.S. intelligence agencies' investigations of domestic suspects, and the disconnect between intelligence and law enforcement. Resolving these problems without unduly infringing on Americans' civil liberties will take judicious new legislation, as well as a restructuring of the executive branch. President Bush's creation of an Office of Homeland Security provides a useful basis for the necessary changes.

At the same time, Washington should pursue an aggressive campaign against the bases of terrorist groups and their possible state sponsors. Terrorist groups often have activities and support scattered in several countries, so the United States needs joint intelligence collection and analysis efforts with other nations, particularly those where terrorist cells are located. It will be perhaps most important (and most difficult) to get this type of cooperation from Russia and China. But just as future success in preventing nuclear proliferation will require joint programs with Russia and China, so will success in collecting intelligence on multinational terrorist groups.

Hostile nations also can pose a danger if they develop the capability to attack the United States with nuclear or biological weapons. In addition to the covert means available to terrorists, states could place their weapons in aircraft, perhaps in the guise of commercial planes, or cruise missiles, perhaps based in freighters off the U.S. coast. Here again, intelligence is key: putting the necessary defense measures in place requires a timely warning of the time and location of a planned strike.

A hostile nation might also strike with long-range ballistic missiles, a possibility that has received a great deal of attention recently. Not wanting to depend on deterrence alone in such a situation, the Bush administration has stated its intent to deploy a national missile defense (NMD) for added protection. NMD is, in a sense, an insurance policy that becomes relevant if both prevention and deterrence fail and the aggressor nation chooses to deliver its weapons using ballistic missiles instead of aircraft, cruise missiles, or covert means. The controversy surrounding missile defense may be thought of as a debate about how likely the United States is to need such insurance, how much the policy will cost, and whether the nation can collect on it if needed (i.e., whether the defenses will work). These are all reasonable questions to ask before committing to the purchase.

The ground-based missile defense system now well advanced in its development is designed to intercept incoming warheads in mid-flight—essentially trying to "hit a bullet with a bullet." Much controversy has arisen about this system, particularly after several test failures. But even though success will demand quite advanced technology, I believe that the United States will demonstrate a convincing mastery of the system before long, perhaps in another five to ten tests. In a few years, therefore, NMD could demonstrate on the test range a technical effectiveness of 80-90 percent.

Assessing the likely operational effectiveness of such a system is a different matter, and it involves taking a realistic view of various possible degrading effects. An NMD system could sit unchallenged for years, for example, and then have to operate perfectly the first time it is needed, probably without any advance warning. Such a scenario is exactly the opposite of the situation on the test range, where the crew is primed and ready (and the firing is postponed if they are not). Experience with other military systems, moreover, suggests that they achieve their best performance only after significant use in combat conditions. Tactical air defenses are fine-tuned after operating against repeated waves of bombing attacks. A missile defense system operating against a nuclear attack would have to perform well during its first and only mission.

In a real attack, finally, one must expect the aggressors to employ technical or tactical countermeasures, such as decoys, chaff, radar jamming, or nuclear-induced radar blackout, to evade the NMD system. Washington is not likely to know which countermeasures might actually be used against its system, but it is prudent to expect them to be tailored to the specifics of the U.S. NMD program as it is described in the public record. Countermeasures are not simple to develop, but the incentive for the missile designer to acquire them is quite high. This inherent vulnerability of an air defense or missile defense system is a problem that can be addressed but never fully resolved.

Susceptibility to countermeasures is not new; indeed, it is a classic weakness common to all air defense systems. Missile defense systems have no significant operational history yet, but the United States and other countries have a history of air defense operations that extends over 60 years. Historically, these activities have demonstrated an ability in combat to shoot down between 3 and 30 percent of an attacking force; under some operational conditions they have done even less well. This record does not stop the United States from building and deploying such air defense systems to defend its military forces from repeated attacks by conventionally armed bombers, because a shoot-down rate even as low as 10 percent would eventually exhaust an enemy's bomber force. But this low success rate is one reason the country has no comparable air defense system capable of defending its cities against a strike from nuclear-armed bombers, for which a shoot-down rate of even 30 percent would be insufficient.

Early in the Cold War, the United States considered deploying an air defense system to protect its population from the growing Soviet bomber fleet. The plans called for large radars, a nationwide command-and-control system, F-106 interceptor planes, and substantial complexes of antiaircraft missiles around each major urban area. A few of these units were actually deployed, but in the end Washington concluded that even if the system could achieve historically high shoot-down rates, it could not provide meaningful national protection against a nuclear attack from the air.

Moscow, meanwhile, made the opposite decision. At the time, U.S. intelligence estimated that to protect their cities against our B-47 and B-52 bombers, the Soviets spent more than $100 billion (in 1970 dollars) building and deploying their air defense system, which included thousands of surface-to-air missiles. In response, the U.S. Strategic Air Command developed technical and tactical countermeasures that they judged would enable a sufficient number of American bombers to penetrate the Soviet defenses and devastate the Soviet Union. This judgment was never put to a test, but its assessment of the Soviet system's vulnerability achieved credibility in the 1980s when a light civilian plane flew from West Germany and landed in Red Square without being intercepted.

The comparison between air defense and ballistic missile defense is imperfect, and one cannot simply apply the track records of the former to the latter. But it is hard to make a persuasive argument that shooting down a ballistic missile is easier than shooting down an airplane, or that a nation capable of deploying a force of intercontinental ballistic missiles could not build relatively challenging countermeasures. Even if the current NMD system eventually demonstrated a 90 percent rate of technical effectiveness on the test range, it is reasonable to question whether it could ever come close to that under operational conditions.

Today's U.S. policymakers must understand the fundamental limitations of missile defense systems against nuclear-armed missiles (just as their predecessors came to understand the limitations of air defense systems against nuclear-armed bombers) and recognize that even if successful in that arena they would provide virtually no protection against a cruise missile or bomber attack, not to mention covert delivery by other means. Failure to recognize these limitations could create a false sense of security and lead to inappropriate defense priorities. In the 1930s, the Maginot line, erected to protect France against a German invasion, had just this effect on French leaders, with terrible consequences for their nation. The Maginot line failed not because it was poorly designed or implemented, but because the Germans recognized precisely how formidable a defense it was and devised a strategy for going around it. Committing the bulk of U.S. homeland defense resources and energies to NMD tempts a similar fate: hostile nations have not only countermeasure options but also the options of carrying their weapons on aircraft or cruise missiles, thus going around our defenses.


Several different NMD systems for protecting American military forces are in advanced stages of development. Theater defenses, which operate against medium-range missiles, will likely be deployed in the next few years at a cost that is reasonably well known. Coming up with a credible estimate of what a national missile defense would cost, on the other hand, is more difficult.

The Bush administration has not yet decided on a final design for such a system, but it has testified that it wants to move to a "layered" approach, in which different components could operate in sequence against a ballistic missile in its boost phase, in midcourse, and in its terminal phase. The Congressional Budget Office has estimated that a full-scale version of the midcourse system now under development would cost $50 billion (for production and deployment of the sites and ten years of operation), plus an additional $10 billion for the space-based sensors. A system aimed solely at a missile's terminal stage probably would not require a new development program, since it could be based on the theater missile defense systems that will be deployed in a few years. But because terminal defenses can protect only relatively limited areas, a national network of them would have to include not only a global command-and-control system but separate and complex packages of missiles and radars for each urban area to be covered.

The boost-phase component of the project has not yet been designed. An air-based version of it, if alerted and deployed in a crisis, might provide an emergency defense against a missile launched from North Korea or Iraq, but not against one from northern Iran. Complete coverage would require either a constellation of spacecraft or bases on the territories of Russia and several of the Central Asian republics. Because of their access to the missile during its boost phase, space-based systems have inherent advantages over those based on the ground, in the air, or at sea. But at the same time, they entail considerably more complex technical problems, raising difficult questions about cost, schedule, and feasibility.

It is hard to imagine either a space-based boost-phase system or a nationwide complex of terminal systems costing less than the ground-based midcourse defense system now under development. In the end, the cost of a layered approach to NMD could be several times higher than the $60 billion estimate for the midcourse system alone—enough to drain significant resources from other military needs. Even if the Defense Department were to save the money it hopes to by reforming the defense acquisition system and closing unnecessary bases, and even with the new willingness since September 11 to commit additional resources to national defense, the administration will have to make difficult choices about how to distribute its spending among force structure, readiness, and new investments, including missile defense.

During my tenure as secretary of defense, I found that setting funding priorities for defense programs and then defending those priorities to the president, Congress, and the public was a very demanding task. I judged then—and continue to believe now—that although the NMD program is important, it should have a lower priority than those programs that are key to maintaining military readiness. I would also accord NMD a lower priority than critical programs designed to upgrade American conventional forces. In particular, I believe there is an urgent need to replace U.S. fighter-bombers with the new generation of aircraft that have been developed over the last ten years, the technology of which (especially stealth capabilities and precise weapon delivery) will give the United States air supremacy in any military conflict for several decades to come. Operation Desert Storm demonstrated how air supremacy enhances all aspects of military operations, allowing the United States to win quickly, decisively, and with minimal casualties. It also illustrated to the rest of the world the futility of directly confronting the U.S. military. The current crisis has once again shown the unique role played by aircraft carriers in rapidly projecting American military power. Washington must support the programs under way to modernize U.S. carrier battle groups. U.S. forces must be transformed with modern information technology, which the Bush administration has rightly made a priority.

All these programs will be expensive, and they will compete with NMD for funding. Unlike NMD, however, these other investments serve more than one purpose. They allow the country to prevail in likely conflicts, they help sustain U.S. global leadership, they help deter conventional war, and they are a vital complement to nuclear forces in deterring the use of nuclear and biological weapons against the United States or its allies. Sacrificing the maintenance of U.S. conventional military supremacy to carry out an extensive NMD program would decrease rather than increase the nation's ability to deter nuclear as well as conventional war.


Responding to the dangers of proliferation and terrorism involves more than defense programs. The United States must also assign a higher priority and devote more funding to intelligence and law-enforcement programs that could help the authorities penetrate those terrorist groups planning attacks, as well as to intelligence efforts that illuminate the nature of the proliferation threat more generally.

Because even the best intelligence efforts can never offer perfect protection, the country also needs to increase its investments in programs designed to cope with an attack once it has occurred. In the case of a biological weapon, for example, quick and effective "consequence management" could reduce prospective fatalities as much as tenfold. Local and state governments, especially firefighters and police, will necessarily be on the front line, but national guard and reserve units can and should be strengthened to provide more effective support. None of these forces, however, has the special equipment, medicine, and training needed to deal fully with a biological attack—only the Centers for Disease Control can direct an effective response. To prepare itself to deal with the wide variety of microbes that might be used in a future attack, Washington must begin immediately to mobilize the medical and pharmaceutical industries so that they will be ready to respond with the needed vaccines, medicine, and health care facilities. All these steps and more, presumably, will be the responsibility of the new Office of Homeland Security, but overcoming bureaucratic divisions and programmatic inertia will be more difficult than some might expect.

Increased efforts to stop or slow proliferation, meanwhile, hold more promise than many critics seem to think. Since the end of the Cold War, four nations (Ukraine, Kazakhstan, Belarus, and South Africa) have given up their sizable nuclear arsenals and two others (Argentina and Brazil) have terminated their nuclear weapons programs short of success, a trend partially offset by the decisions of India and Pakistan to come out of the nuclear closet. Continuing the existing nonproliferation efforts is important, as is aggressively pursuing opportunities to reduce new threats before they emerge—for example, by negotiating an agreement whereby North Korea abandons its long-range missiles.

Serious nonproliferation efforts must involve Russia and China. Sustained dialogue with both is crucial, but the most important subject for such dialogue is proliferation, not missile defense or even reductions in strategic forces. Moscow and Beijing must take serious actions, in cooperation with the United States, to curb the unconventional weapons programs in Iraq, Iran, Libya, and North Korea. To get Moscow and Beijing on board, Washington should be prepared to make some compromises on other issues. Both governments appear to care less about proliferation than about preserving their ultimate nuclear deterrent. The United States should take the opposite approach, thus opening space for mutually beneficial discussions.

If effective agreements to curb proliferation cannot be reached, the threat will continue to grow. Indeed, if the present impasse in the consultations on missile defense continues, it could lead China to dramatically increase the long-range missile modernization program it now has under way and could lead both Russia and China to provide missile and counterdefense technology to nations hostile to the United States. If the attempt to deploy a missile defense resulted in an increase in proliferation, it would represent a net decrease in U.S. security. If discussions with Russia and China could succeed in reaching meaningful proliferation curbs, on the other hand, the Bush administration would seize a unique and historic opportunity to prevent new nuclear and biological threats from emerging. It is of course possible that the needed cooperation will not be forthcoming from Russia or China. But the stakes are too high to not make every effort.

If the Bush administration works to maintain U.S. conventional military supremacy, boosts efforts at intelligence gathering and consequence management, and pursues international cooperation on the pivotal nonproliferation issue, it is unlikely to have enough funds or diplomatic leverage for the near-term deployment of a full-scale, layered NMD system. It should still be possible, however, to support an accelerated program to produce and deploy theater missile defenses. Once the new systems have been developed, they could be deployed rapidly during crises to defend against ballistic missile threats in those (relatively few) cases where the missile's boost phase would fall within range of the system. Deploying a naval-based missile defense system or an airborne laser to South Korea, for example—as the Clinton administration deployed Patriot missiles during the 1994 crisis caused by the breakdown in nonproliferation talks with North Korea—would be one way to respond to an attempt at nuclear extortion.

It should also be possible to maintain a robust missile defense research and development program, the results of which might change the calculus on such issues down the road. A central objective of this effort should be to gain a sophisticated understanding of missile defenses' vulnerability to countermeasures and develop appropriate means to defeat those countermeasures. In particular, the Defense Department should have a much more aggressive program to test the performance of American NMD systems against all realistic countermeasures. Testing can play an important role in validating the design of mechanisms to thwart countermeasures. But only very detailed and extensive simulations, monitored by an objective "red team" of outside observers, can allow officials to evaluate how well the system would work against the diverse countermeasures that it might have to face.


The United States has suffered the most devastating terrorist attack in world history. It can and will respond. But the attack demonstrates that there are large, well-organized groups whose primary objective is to kill large numbers of Americans. These groups understand all too well that nuclear or biological weapons can fulfill that mission even better than truck bombs or kamikaze aircraft can. The United States also faces a small number of nations that believe they can advance their own interests by mounting unconventional threats. Future U.S. security therefore depends on actions taken today to prevent the proliferation of nuclear and biological weapons tomorrow. Nonproliferation efforts, in turn, depend on effective cooperation with the other nuclear powers. Achieving such cooperation is therefore a critical national security objective.

Even if the United States fails to prevent proliferation, it still has a powerful and credible deterrent of both nuclear and conventional weapons. But it is reasonable to take out insurance against the contingency that both prevention and deterrence fail. National missile defense is such an insurance policy. As the government considers the priority to give to missile defense relative to other national security efforts, both within the defense budget and without, it should recognize that NMD would not provide any protection against the most likely forms of terrorist attack, nor would it be effective against a strike by cruise missiles or bombers. The insurance policy would thus cover a possible but not the most likely contingency, would come at a high price, and could stimulate an increase in the level and sophistication of the threats the country faces.

Theater missile defenses, in contrast, address a clear and direct threat to American deployed forces from short-range missiles, and the military should move to deploy the next generation of them as expeditiously as possible. It makes sense to continue a robust research and development program for defenses against ballistic missiles, but it would be a mistake to let such efforts interfere with attempts to prevent proliferation or hamper achieving the joint international programs necessary to respond effectively to the immediate terrorist challenge. In any event, informed judgments about the wisdom of deploying an NMD system can be made only after officials can get realistic estimates of its effectiveness in the face of probable countermeasures and credible estimates of its financial and diplomatic costs. That day is still several years away.

You are reading a free article.

Subscribe to Foreign Affairs to get unlimited access.

  • Paywall-free reading of new articles and a century of archives
  • Unlock access to iOS/Android apps to save editions for offline reading
  • Six issues a year in print, online, and audio editions
Subscribe Now
  • William J. Perry, Professor in the Department of Management Science and Engineering at Stanford University, served as U.S. Secretary of Defense from 1994 to 1997.
  • More By William J. Perry