In the domain of national security, many people favor some kind of precautionary principle, insisting that it is far better to be safe than sorry and contending that a range of important safeguards, including widespread surveillance, is amply justified to prevent the loss of life. Those who object to the resulting initiatives, and in particular to widespread surveillance, respond with a precautionary principle of their own, seeking safeguards against what they see as unacceptable risks to privacy and liberty. The problem is that, as in the environmental context, a precautionary principle threatens to create an unduly narrow viewscreen, focusing people on a mere subset of the risks at stake. What is needed is a principle of risk management, typically based on some form of cost-benefit balancing. For some problems in the area of national security, however, it is difficult to specify either costs or benefits, creating a severe epistemic difficulty. Considerable progress can nonetheless be made with the assistance of four ideas, calling for (1) break-even analysis; (2) the avoidance of gratuitous costs (economic or otherwise); (3) a prohibition on the invocation or use of illicit grounds (such as punishment of free speech or prying into people’s private lives); and (4) maximin, which counsels in favor of eliminating or reducing the risk of the very worst of the worst-case scenarios. In the face of incommensurable goods, however, the idea of maximin faces particular challenges.