"This dissertation examines how weak states manipulate the incentives of strong states to achieve their own ends in international bargaining over alliances and international cooperation in areas such as counter-terror operations, nuclear (non-)proliferation, and conflict. First, I examine moral hazard problems in international aid-for-policy relationships. Donor states face a compliance dilemma. On the one hand, providing aid generates short-term incentives for a client state to comply with desired policies such as eliminating terrorists within their borders. On the other hand, the client state may "fail by succeeding"- failure to eliminate terrorists creates incentives for the donor to continue to provide aid in the future. I present a formal moral hazard model to study the compliance dilemma. It shows that the shadow of the future is sufficient to cause suboptimal efforts to fight terrorists in the client state, even if their government receives a direct benefit from monopolizing violence. One partial solution is for donor states to launch a limited intervention independent of their allies. This intervention directly contributes to achieving their objectives but also reduces the bite of the moral hazard issue. Examining the relationship between the United States and Pakistan illustrates the mechanisms of the model. Second, I examine how nuclear latency development - the ability to build nuclear weapons - affects a state's ability to secure a place under the U.S. nuclear umbrella. The United States has long opposed nuclear proliferation. However, empirically, many U.S. allies have chosen to research and develop nuclear capabilities. Existing theories do not account for the interrelated strategic choices to enter alliances and pursue nuclear weapons, nor do empirical estimates incorporate this strategic behavior. I construct a strategic probit model based on a new theory in which weaker states strategically manipulate nuclear development to influence strong states' alliance offers. Strong states select weak states into alliances to change their nuclear development calculus. The estimates overturn two pieces of conventional wisdom. First, nuclear latency can push past entrapment concerns in uncertain security environments to drive alliance formation. Second, I show that high latency states that are militarily strong can use latency to secure alliance guarantees. Finally, I examine the effect of alliances with a militarily strong state on the risk of protege-initiated conflict. The literature widely argues that asymmetric alliances, where a powerful, often nuclear-armed state commits to the defense of a weaker protege state, encourages revisionism against third parties. The leading explanation for this tendency is moral hazard. However, in contrast to this literature, I argue that adverse selection, driven by uncertainty over patron preferences at the time of alliance formation, drives both the pattern of conflict initiation observed in the literature and significant restraint by proteges in alliances. I build a formal model of asymmetric alliance formation and coercive bargaining and then test the theory using an empirical strategic model of alliance formation that allows for adverse selection into alliance arrangements. Once I account for the adverse selection problem, I find that alliances are restraining."--Pages vii-viii.