Online Debate

PrintPrint EmailEmail ShareShare CiteCite


How Likely is a Nuclear Terrorist Attack on the United States?

Discussants: Michael A. Levi, David M. Rubenstein Senior Fellow for Energy and the Environment and Director of the Program on Energy Security and Climate Change, and Graham T. Allison, Director, Belfer Center for Science and International Affairs, Harvard Kennedy School
Updated: April 20, 2007

U.S. policymakers agree that as possible terrorist attacks go, the worst-case-scenario would involve detonation of a nuclear bomb in a major American city. This most catastrophic of scenarios provides ample fodder for the plot of television dramas, but the actual likelihood of such an event is open to debate.

Michael A. Levi CFR Fellow for Science and Technology and author of the forthcoming On Nuclear Terrorism and Graham T. Allison, director of Harvard's Belfer Center for Science and International Affairs and author of Nuclear Terrorism: The Ultimate Preventable Catastrophe, consider the odds of such a devastating attack

Weigh in on this debate by emailing the editors at To view other online debates click here.

Graham T. AllisonMost Recent

April 20, 2007

Graham T. Allison

A final comment on the likelihood of a nuclear terrorist attack before turning more specifically to terrorist motivations.

We should ask ourselves every day: Are nuclear materials that could fuel a terrorist's bomb more or less secure than they were a year ago? Thanks to initiatives like the Nunn-Lugar program, highly enriched uranium and plutonium in Russia are far safer from theft today than they were in the early 1990s. But the risk that terrorists will buy or steal nuclear material from a rogue state increases as more countries acquire the ability to produce weapons-usable material. Therefore it is vitally important to roll back North Korea's nuclear program and to constrain Iran before it reaches its enrichment finish line. By becoming a nuclear-armed state, each will trigger a cascade of proliferation in its neighborhood.

What about the motivation of terrorists that have attacked the American homeland? Al-Qaeda spokesman Suleiman Abu Gheith has stated al-Qaeda's objective: "to kill 4 million Americans—2 million of them children—and to exile twice as many and wound and cripple hundreds of thousands." As he explains, this is what justice requires to balance the scales for casualties supposedly inflicted on Muslims by the United States and Israel. Michael Levi argues, correctly, that such a tally could be reached in a series of smaller installments, and our national security would benefit from insights into how to prevent such events. But ask yourself how many 9/11s it would take to reach that goal. Answer: 1,334, or one nuclear weapon.

Jihadi terrorists are not solely interested in murdering Americans. They are also vying for Muslim "hearts and minds" by demonstrating that al-Qaeda is the "strong horse." Bin Laden has challenged his followers to trump 9/11. The London and Madrid train bombings set a bar: the first major bombing by Islamic terrorists on each country's soil. Al-Qaeda's next UK plot was more audacious, and had it been successful, it would have taken more lives.

It is not clear that al-Qaeda can be deterred. Osama bin Laden describes the current conflict as a clash between the Muslim ummah [community of believers] and the "Jewish-Christian crusaders." A nuclear terrorist attack, like the bombing of Hiroshima and Nagasaki, would be a world-changing event. Bin Laden well might accept significant risk of failure for a chance to draw battle lines in his clash of civilizations.

Analysts with a deeper understanding of terrorist motivations should be challenged to propose policy initiatives that leverage that knowledge, particularly where those insights help us to prevent what Dr. Levi and I both agree would be the single greatest catastrophe: nuclear terrorism.

Michael A. LeviMost Recent

April 19, 2007

Michael A. Levi

Let me begin as I did last time by agreeing that a low probability of a nuclear terrorist attack is something that we must treat with more urgency than our actions have reflected so far. It seems that we more than agree on this front.

I'll focus again on the question of probability. I am wary of taking "Dragonfire" at his word about the likely yield of a terrorist bomb, given that essentially everything else he said turned out to be a lie. No doubt a terrorist bomb might have a ten kiloton yield. But it might also have a much smaller one. In contrast with what Dr. Allison implies, a "fizzle" yield for the simplest type of nuclear weapon (a gun-type bomb as he rightly pointed out in his last post) is likely to be roughly ten tons, not one kiloton.

Should this observation change the urgency with which we confront nuclear terrorism? Not a whit. As Dr. Allison will surely agree, even if such a bomb had far more limited physical effects than a ten kiloton device, its consequences would still be horrific. The prospect of mass destruction on a smaller scale is still something that should spur us to stronger action.

Yet from a terrorist perspective the prospect of a fizzle or a dud might change things. Let me start by revisiting the question of terrorist aversion to failure—terrorist motivations are central to the likelihood of nuclear terrorism, and we seem to disagree on what they are. I have never asserted that terrorists will not attempt anything but "foolproof" plots. But there is a lot of territory in between foolproof and a 90 percent (or even 50 percent or 30 percent) chance of failure.

Why might a group decide against a course of action with a 10 percent chance of killing tens or hundreds of thousands? A group might have better alternatives. An attack on public transportation that has a ninety-five percent chance of killing forty people is a straw man alternative to nuclear terrorism—certainly terrorist groups have intermediate and perhaps, from their perspectives, more compelling options, like suicide aircraft attacks, Madrid and London style bombings, and plots like the one using liquid explosives that failed last summer.

Here is another possibility: In the wake of a full-blown nuclear plot, the international campaign against terrorism would likely step into a much higher gear. Would al-Qaeda accept a ninety percent chance of failing to kill more than a massive conventional bomb would while incurring a large risk of provoking a response that might cripple its ability to initiate other plots, nuclear or non-nuclear, in the future? We can't know the answer, but there is no reason to assume that al-Qaeda would choose such a course.

If we ignore the possibility that terrorists will be dissuaded by relatively small risks of failure, we are likely to dismiss a host of limited defensive options that might otherwise substantially lower the likelihood of nuclear terrorism. Rather than demanding a perfect defense—something that is unachievable—we should leverage what we know about terrorist psychology to minimize the odds of catastrophe.

Graham T. AllisonApril 18, 2007

Graham T. Allison

Let's run a little with Michael Levi's numbers. Imagine that he is correct, and terrorists have "a 90 percent chance of failure" if they attempt a nuclear 9/11. On the flip side, that would mean a 10 percent chance of success. What should a 10 percent possibility of success mean in terms of U.S. policy? Remember, risk equals probability times consequences.

On a normal workday, half a million people crowd the area within a half-mile radius of New York City's Times Square. If, in the heart of midtown Manhattan, terrorists detonated a ten-kiloton nuclear bomb (the yield of the bomb an intelligence source codenamed "Dragonfire" claimed was in New York one month after 9/11), the blast would kill them all instantly. Hundreds of thousands of others would die from collapsing buildings, fire, and fallout in the hours and days thereafter.

Multiply the consequence of such an attack (five-hundred thousand souls) by a 10 percent probability, and one would conclude that the U.S. government should mobilize an effort to prevent nuclear terrorism equivalent to saving fifty thousand Americans lives.

Furthermore, the effect of a nuclear terrorist attack would reverberate beyond U.S. shores. After a nuclear detonation, the immediate reaction would be to block all entry points to prevent another bomb from reaching its target. Vital markets for international products would disappear, and closely linked financial markets would crash. Researchers at RAND, a U.S. government-funded think tank, estimated that a nuclear explosion at the Port of Los Angeles would cause immediate costs worldwide of more than $1 trillion and that shutting down U.S. ports would cut world trade by 7.5 percent.

But Dr. Levi raises the possibility that, were terrorists to get their hands on enough nuclear weapons material to make a bomb, their design might fail. If a terrorist's ten-kiloton nuclear warhead were to misfire (known to nuclear scientists as a "fizzle") and produce a one-kiloton blast, bystanders near ground zero would not know the difference. Such an explosion would torch anyone one-tenth of a mile from the epicenter, and topple buildings up to one-third of a mile out.

Does the real possibility of a fizzle or failure mean that terrorists won't attempt a nuclear attack? Not necessarily. If terrorists pursued only fool-proof plans, they would have begun suicide bombing attacks on U.S. public transportation by now. But from a terrorist's point of view, why pursue a course of action with a 95 percent chance of success, but at most forty victims, if you have a 10 percent chance at killing five-hundred thousand?

The most important takeaway from this debate is that we must do everything technically feasible on the fastest possible time line to prevent terrorists from getting their hands on nuclear materials.[1] Whether nuclear explosion, fizzle, or total dud, the repercussions of such materials in jihadist clutches are unacceptable.

[1] Policy solutions to this end are developed in detail in Nuclear Terrorism: The Ultimate Preventable Catastrophe (New York: Times Books, 2004), chapters 7 and 8.

Michael A. LeviApril 17, 2007

Michael A. Levi

Let me start with some points of agreement. Al-Qaeda has grand ambitions and seeks mass casualties. And regardless of the probability of nuclear terrorism, the potential consequences of a successful attack should be enough to prompt us to more urgent action than we are currently taking.

We should not, however, underestimate the odds of terrorist failure. There isn't enough space here to make that point comprehensively, but I'll try to convince you that simple arguments for why failure is highly unlikely may be weaker than they seem.

The case for the ease of building a gun-type weapon provides a good example of how we often overestimate how easy a terrorist task may be. I certainly won't debate the fact that Manhattan Project scientists "were so confident about this design that they persuaded military authorities to drop the bomb, untested, on Hiroshima." But we should parse the word "untested" carefully.

During the Manhattan Project, scientists and engineers spent years testing the gun itself; testing their casting and machining of the uranium metal to avoid fires and criticality accidents during production, and impurities in the product; testing the initiator that would trigger the chain reaction; and testing how different configurations of materials would behave, a project that led to the death of one physicist. No one conducted a full-scale test explosion, but that hardly means that building the weapon was trivial.

A terrorist group would have to do many of the same things (though technological progress would make some steps easier) all while attempting to hide from law enforcement and intelligence. This doesn't mean that terrorists couldn't build a gun-type bomb, but it suggests that their chances of failure aren't negligible.

This takes on special importance in the context of a broader defense. Imagine a terrorist group faces only a twenty percent chance of failure while building a bomb. But imagine it also faces a similarly small chance of failure while attempting to purchase nuclear materials, while attempting to recruit scientists and engineers, while raising money for its plot, while smuggling materials into the United States, while purchasing non-nuclear components for its weapon, while assembling the bomb in a safehouse, and in other elements of its plot. If we combine, for example, ten such hurdles, we get a ninety percent chance of failure. We can debate the numbers, but this suggests that we shouldn't be too quick to ignore small chances of terrorist failure.

A final note on the question of failure-aversion, a quality most terrorism analysts still, even after 9/11, attribute to most terrorist groups. I agree that al-Qaeda is patient and plans carefully. But that does not mean that after careful and methodical consideration, and facing a properly designed defense, al-Qaeda might not decide that a nuclear plot is too much of a stretch to seriously try.

Graham T. AllisonApril 16, 2007

Graham T. Allison

In the hotly contested American presidential election in 2004, the two candidates agreed on only one fundamental point. In the first televised debate, they were asked, what is "the single most serious threat to the national security to the United States?" President Bush, answering second, said: "I agree with my opponent that the biggest threat facing this country is weapons of mass destruction in the hands of a terrorist network."

I also agree. This debate asks how likely is it that terrorists will explode a nuclear bomb and devastate a great American metropolis. In the judgment of former U.S. Senator Sam Nunn, the likelihood of a single nuclear bomb exploding in a single city is greater today than at the height of the Cold War. Nuclear Terrorism states my own judgment that, on the current trend line, the chances of a nuclear terrorist attack in the next decade are greater than 50 percent. Former Secretary of Defense William Perry has expressed his own view that Nuclear Terrorism underestimates the risk.

From the technical side, Richard Garwin, a designer of the hydrogen bomb who Enrico Fermi once called, "the only true genius I had ever met," told Congress in March that he estimated a "20 percent per year probability with American cities and European cities included" of "a nuclear explosion—not just a contamination, dirty bomb—a nuclear explosion." My Harvard colleague Matthew Bunn has created a probability model in the Annals of the American Academy of Political and Social Science that estimates the probability of a nuclear terrorist attack over a ten-year period to be 29 percent—identical to the average estimate from a poll of security experts commissioned by Senator Richard Lugar in 2005.

Rather than quibble over percentage points, the bottom line is recognition that risk equals probability times consequences. Even skeptics who believe that experts overestimate the probability find it difficult to discount the risk.

Prior to 9/11, most terrorism experts argued that terrorists sought not mass casualties but rather mass sympathy through limited attacks that called attention to their cause. But after that horrific attack, the bipartisan 9/11 Commission issued its major conclusion: The principal failure to act to prevent the September 11 attack was a "failure of imagination." A similar failure of imagination leads many today to discount the risk of a nuclear 9/11.

It is a mistake to confuse al-Qaeda's patience and careful planning with the view that they "hate to fail" or lack grander ambitions. When Khalid Sheikh Mohammed, the chief planner of 9/11, first proposed an easier plan to charter a small plane, fill it with explosives, and crash it into CIA headquarters in Langley, Virginia, Osama bin Laden replied, "Why do you use an axe when you can use a bulldozer?"

Finally, a crude gun-type bomb built from highly enriched uranium would be relatively simple to construct and reliable. Manhattan Project scientists were so confident about this design that they persuaded military authorities to drop the bomb, untested, on Hiroshima.

Michael A. LeviApril 13, 2007

Michael A. Levi

How likely is a nuclear terrorist attack on the United States? I doubt anyone knows. I also suspect that the exact answer isn't all that important—even a small chance of catastrophe is worth worrying about. What is valuable is thinking through those factors that make nuclear terrorism either more or less likely; even if the exercise doesn't yield a definitive conclusion, it helps us figure out how to prevent nuclear terrorism. I want to argue that many analyses of nuclear terrorism miss important factors that tend to make nuclear terrorism less likely than it otherwise would be. Today I want to highlight one: terrorists' fears of failure.

A nuclear weapon requires highly enriched uranium (HEU) or plutonium, materials that don't occur in nature and that terrorist groups cannot produce themselves. The ease of access to materials in state stockpiles is thus one of the main factors affecting the odds of a nuclear terrorist attack. The other big factor is motivation. Most terrorist groups have little incentive to pursue nuclear terrorism, since mass murder doesn't serve their political ends—but for some groups, indiscriminate killing is precisely the goal. Most analysts agree that the availability of nuclear weapons and materials, and the utility to terrorist groups of successful nuclear attacks, are the two most important factors in determining the likelihood of nuclear terrorism, even if they disagree over how hard acquiring materials would be or over how many groups might expect to benefit from nuclear terrorism.

So let me flag another dimension of motivation that gets too little attention. Even groups that want to and possibly can execute nuclear attacks may decide against them. Why? Because many of the most dangerous terrorist groups hate to fail. Brian Jenkins wrote recently that for jihadists, "failure signals God's disapproval." That's a lot of pressure to succeed. This inevitably pushes the odds of nuclear terrorism down. When we look at our defenses against nuclear terrorism, we prudently notice the holes. When terrorists look at those same defenses, they may be fixating on whatever barriers, however limited, exist. If that's what's happening, nuclear terrorism may be much less likely than many expect.

More on This Topic