Abstract: Symposium: Free Speech in War Time Conference. In a standard analysis, the history of civil liberties is characterized by a series of security panics. A range of mechanisms - cognitive heuristics and biases, various forms of cascading and herding, conformity and preference falsification, and so on - cause periodic panics in which aroused publics demand repressive measures to curtail the civil liberties of perceived enemies of the nation, particularly noncitizens or other outsiders. Government officials may themselves panic, or will at least supply the panicky measures that constituents demand. The standard remedy is to urge changes to legal doctrine or institutions, in order to curtail government's power to repress civil liberties in response to security panics. The standard model of security panics has been criticized on several grounds. Sometimes security panics are justified, even if produced by disreputable mechanisms; fear can motivate beneficial action as well as detrimental action. In any event, legal doctrines, and perhaps even institutional design, will prove incapable of constraining a genuinely panicked public. In what follows I will sketch a different criticism of the standard model. Even if that model is right as far as it goes, it is fatally incomplete. My central claim is that the mechanisms underlying security panics have no necessary or inherent pro-security valence. The very same mechanisms are equally capable of producing libertarian panics: episodes in which aroused publics become irrationally convinced that justified security measures represent unjustified attempts to curtail civil liberties. I will suggest that libertarian panics have been a regular occurrence in American history, and that we may be living through one now, in the form of a widespread and thoroughly irrational, even hysterical, reaction to small legal changes adopted after 9/11. Indeed, the tendency to diagnose the existence of a security panic can itself be symptomatic of a libertarian panic. The existence of libertarian panics undermines the institutional reforms urged by the advocates of the standard model. The very reforms that would minimize the risks and harms of security panics will maximize the risks and harms of libertarian panics. The institutional-design problem, then, is to optimize in light of these offsetting risks; whatever legal and institutional arrangements turn out to be optimal, they will necessarily prove less protective of civil liberties than the arrangements favored by advocates of the standard model.