This dissertation contains two major themes: providing an operational definition for regime recognition and constructing a dynamic theory of regime deviation. Through the application of formal, behavioral, and cognitive approach of regime recognition, the case of the trilateral relationship between the United States, China, and, Taiwan on the issue of the cross-Strait dispute since 1972 to 2000 was identified as an international regime. To create a dynamic theory of regime deviation, I assume that national interest is a variable dependent on the variation of domestic preferences distribution and that the paramount leader and government are pivotal players and expected utility maximizers trying to satisfy demands and pressures from the level of the international system and domestic constituency under the governance of an international regime. In an effort to construct a concise framework of regime robustness focusing specifically on the levels of individual and domestic society, I put forward two models of regime robustness for democratic and nondemocratic states based on Putnam's two-level game concept and proposed four hypotheses about when and how the paramount leader and government would respond to disgruntled domestic forces and under what circumstances the paramount leader and government would take drastic deviation from the order of a regime. After the analysis of the three cases of regime deviation, the proposed models and hypotheses are proved feasible and valid. It provides us a better understanding of the dynamics of preference change, the role of leaders, and the politics of domestic, constituency-driven regime deviation under the governance of an international regime.