The future of civilization, perhaps of the human race, hinges on our ability to avoid nuclear war. This point has been made so repeatedly and occasionally so eloquently that it needs no amplification here (Dyson, 1984; Katz, 1982; Institute of Medicine, 1986; Sagan, 1983; Schell, 1980). The consensus, however, begins and ends on this point. There is wide disagreement on how likely nuclear war is, how such a war might occur, the forms it might take, and how it might best be prevented. Will nuclear war arise as the result of a "conflict spiral" between the nuclear superpowers—a self-reinforcing process driven by the tendency of each side to exaggerate the hostile intent of the other and to acquire ever more sophisticated weapons systems that increase the other side's sense of vulnerability and motivation to strike first in a crisis? Will nuclear war arise as a result of the failure of deterrence—a failure to convince the other side that one has both the political will and military capability to resist encroachments on "vital national interests"? Will nuclear war arise as a result of accident or miscommunication triggered by flaws in the command, control, and intelligence systems of the superpowers? Or has the casual role of the superpowers been overestimated? Will nuclear war arise from Third World conflicts of relatively remote relevance to U.S.-Soviet relations? And must nuclear war be an all-or-nothing proposition? Might limited nuclear wars periodically break out between particular powers or combinations of powers?