An experiment was conducted to obtain empirical evidence that controllers, in addition to detecting and resolving expected losses-of-separation, intervene to mitigate potential conflicts that would not result in a loss of separation assuming proper compliance by both pilots. Participant controllers controlled traffic that contained aircraft expected to lose separation as well as others that would not lose separation, but were configured in such a way that a loss of separation could occur quickly if one of the aircraft did not comply with their clearance. Binary logistic regressions performed on the aircraft that were subjected to control by the aircraft provides evidence that this mitigation set improves the predictive power of a model to identify aircraft to which the participants applied control. Moreover, a regression performed on the aircraft for whom controllers provided informative commands, particularly traffic identification calls, demonstrated differences between the loss-of-separation aircraft and the mitigation aircraft. The results have implications for understanding strategies used by air traffic controllers when deciding for what aircraft controllers will intervene.