Graham Allison was among the first scholars to sound the alarm about the risks of Russian loose nukes, and in "How to Stop Nuclear Terrorism" (January/February 2004), he continues to warn of this underappreciated danger. He is right to highlight the inadequacy of current U.S. and international efforts to deny terrorists access to fissile material. And he makes a compelling case for the need to develop a more coherent and multilateral strategy for stopping them.
Unfortunately, Allison's article is less successful in describing the full scope of the problem and recommending a nuanced response. In particular, it fails to distinguish among four distinct types of nuclear terrorism, the relative risks posed by the main two types of fissile materials used in nuclear weapons, and the means necessary for keeping such weapons from nonstate actors as well as states.
OUT OF STATE
Allison addresses what is arguably the most urgent aspect of the threat of nuclear terror: the danger that terrorists will acquire the fissile materials needed to make nuclear bombs. His article, however, conveys the misleading impression that nuclear terrorism is a unitary phenomenon. In fact, terrorists present at least four different kinds of nuclear threats: that they will disperse highly radioactive material by conventional explosives (i.e., "dirty bombs") or other means, that they will attack or sabotage nuclear power installations, that they will seize intact nuclear weapons, and that they will steal or buy fissile material for the purpose of building a nuclear bomb. All four threats are real and merit the attention of policymakers. All four will be expensive to prevent or protect against. They all vary widely, however, in the probability that they will actually occur, in their potential for causing harm, and in the ease with which they can be prevented.
A fundamental shortcoming of the current U.S. policy to combat nuclear terrorism is that it fails to take these differences into account. As a result, Washington has no guidelines for directing its limited