Military arbitrations are cases in which armed forces are ordered by civilian rulers to repress their own populations during domestic contentious events, such as protests and demonstrations. In these cases, a regime essentially forces its military to become involved in sovereign power issues; officers can choose to obey and side with the ruler, or disobey and effectively side with the people. This dissertation proposes and develops a theory of military arbitration to explain armed forces’ frequently surprising decisions to either stand aside and let a ruler fall, or repress their own civilian populations during these volatile periods. It demonstrates the theory’s causal logic at work through process tracing of military arbitrations cases in Serbia, China, and Egypt. Finally, this dissertation examines several existing explanations of military arbitration, highlighting why they are insufficient for explaining military behavior during these unique crisis events.