Security relationships between states can be categorized as more or less hierarchical. Some relationships are characterized by dominance and subordination, while others are highly egalitarian. Why relationships vary along this continuum has never been studied quantitatively in the International Relations literature. In this thesis I set out to test the most popular and well-developed theory on this subject, namely the transaction-cost theory of international security relationships. Using a range of variables and datasets from the quantitative International Relations literature, I have developed a research design to test the empirical implications of the transaction-cost theory on the subject of the design of military alliances. Some alliances are designed with hierarchical safeguards that allow a powerful state to restrict the autonomy of a weaker ally, and the transaction-cost theory should be able to account for these alliances. Using a logistic regression model, I have investigated the effects of transaction-cost variables on the choices of alliance design. I find that most of the hypotheses that are derived from the transaction-cost theory are discarded. Meanwhile, a model that includes successful transaction-cost variables offers significant explanatory and predictive power, and is a substantial improvement on a baseline model of variables derived from more mainstream International Relations theory. My analysis provides new and valuable knowledge about which factors are decisive in pushing the governance of security relationships in a hierarchical direction. It seems that powerful states` fears of being pulled into unwanted conflicts, institutional dissimilarity, previous colonial relationships, and asymmetries in size and material power go a long way in explaining why some states end up in hierarchically organized military alliances.