I argue that conflict can best be understood as a learned practice that constrains states interacting peacefully or conflictually with other states in the international system. Vasquez (1993/2009) proposes that conflict is a learned behavior that emerges from a prior pattern of interactions. And if Senese and Vasquez (2008) are correct that we can understand behaviors such as alliances, arms races, rivalries, territorial disputes, and a non-democratic polity as increasing the probability of conflict, I maintain that we can understand the steps to war as learned behaviors. In other words, I maintain that alliances, arms races, and disputes in the presence of rivalries are learned behaviors. To provide evidence for this claim, I build a matrix of all available alliance texts from 1891-1995 and demonstrate why some states make strategic choices to copy prior alliance texts. Next, I argue that the arms race between the Soviet Union and the United States can be understood as a learning event, as each move by a side during an arms race reveals intentions, motivations, and values. Finally, I argue that evidence of diffusively learned conflict behavior can be found by examining states in rivalries and their interactions with non-rivalrous dyads. I find evidence to support my claim that the behaviors associated with the steps to war argument are learned behaviors.