The purpose of this study is to logically analyze the U.S. extended deterrence policy on the Korean Peninsula by applying game theory. Accordingly, a three-party game model was created with North Korea as the Challenger, South Korea as the Protege, and the United States as the Defender. Considering the uncertainty of international relations, the equilibrium was analyzed under incomplete information where the player does not know the type and behavior of the other player. In particular, in order to focus on the analysis of the reliability of extended deterrence, the type of reliability in the alliance between Korea and the U.S. was set as incomplete information. As a result of the research model analysis, if pS and pU, which are the probability that Korea and the U.S. are alliance reliable types, meet certain conditions, four equilibria could exist: Deterrence 1 and Deterrence 2, Korea Attack 1, and Korea Attack 2. Based on this, the implications were drawn that strengthening conventional forces, possessing nuclear latency, deploying U.S. strategic assets, and promoting nuclear sharing plans could help strengthen extended deterrence.