Sagan informs his readers that he undertook this project believing that the chance of an accidental nuclear explosion was very low, and that he came to a thoroughly different conclusion. The Limits of Safety makes the chilling case that the United States (and, one must assume, the Soviet Union) came close more than once to some very nasty episodes. The book rests on two pillars: a great deal of research in public archives and the persuasive theory of a Yale sociologist, Charles Perrow, whose Normal Accidents makes the case that accidents are not the exception but the rule and that efforts to prevent them sometimes increase their likelihood. Readers will have to make their own judgments about just how dangerous the incidents Sagan describes really were, and the relative paucity of material after the 1960s certainly limits his argument. In the final analysis, no accident occurred that brought the two superpowers close to war. Still, his warnings about the dangers created by the nuclear arsenals of less sophisticated or safety-minded powers should discomfit academic proponents of stability through nuclear proliferation. The findings of The Limits of Safety will surely lend themselves to sensational stories in the press, but that is not the fault of the author, who has written an interesting and troubling book.