States pursue their cooperative and competitive goals using both public and private policy tools. Yet there is a profound mismatch between the depth, variety, and importance of covert activity and what scholars of International Relations (IR) know about it. This article addresses this gap by analyzing how adversaries struggle for influence within the covert sphere, why they often retreat to it, and when they abandon it. It focuses on secrecy among adversaries intervening in local conflicts and develops a theory about secrecy's utility as a device for creating sustainable limits in war. Drawing on insights about secrecy and face-work from the sociologist Erving Goffman, I show that major powers individually and collectively conceal evidence of foreign involvement when the danger of unintended conflict escalation is acute. Doing so creates a kind of “backstage” in which adversaries can exceed limits on war without stimulating hard-to-resist pressure to escalate further. An important payoff of the theory is making sense of puzzling cases of forbearance: even though adversaries often know about their opponent's covert activity, they often abstain from publicizing it. Such “tacit collusion” arises when both sides seek to manage escalation risks even as they compete for power and refuse to capitulate. The article evaluates the theory via several nested cases of external intervention in the Korean War. Drawing on newly available materials documenting the covert air war between secretly deployed Soviet pilots and Western forces, the cases show how adversaries can successfully limit war by concealing activity from outside audiences. Beyond highlighting the promise in studying the covert realm in world politics, the article has important implications for scholarship on coercive bargaining, reputation, state uses of secrecy, and how regime type influences conflict behavior.