Automated emergency maneuvering systems can avoid or reduce the severity of collisions by taking control of a vehicle away from the driver during high-risk situations. To do so, these systems must be capable of making certain decisions: there is the choice of when to interfere in the driver's control, which is made challenging in the presence of dynamic obstacles and uncertain information (such as that provided by imperfect sensors), and there is the choice of how to interfere, since, in general, controls that mitigate collisions will not be unique. We address both of these questions with a probabilistic decision threshold framework that overrides the driver's control only when safety drops beneath a problem-specific threshold, and minimizes the magnitude of the interference. We demonstrate its application to two scenarios: collision-imminent braking for obstacles traveling along the vehicle's path, and braking and accelerating for unprotected lane crossings. © 2012 IEEE.