Early warning systems represent an effective tool for disseminating timely information about potentially catastrophic hazards to the public, emergency managers, and operators and users of transportation systems, industrial facilities, public buildings including schools and government offices, and private residences. This information gives the possibility of taking action to initiate mitigation or security measures before a catastrophic event occurs. Earthquake early warning technologies represent a promising application, especially in areas where faults are located in the proximity of an urbanized area, such as in Los Angeles and San Francisco, California. For earthquake early warning systems, automated procedures for taking action are necessary due to the short lead time available, including an automated decision procedure to decide when to activate mitigation or security measures based on the expected consequences of taking, or not taking, such actions. In this work, a decision procedure is presented that accounts for the uncertainty in the predicted ground shaking intensity at a site of a facility by deciding whether or not to take a mitigation action based on the probabilities of false and missed alarms. It uses a cost-benefit analysis of the consequences of making a wrong decision. It is shown how the method may be extended to other predictors, such as expected losses, that may be more effective parameters for rational decision making from the perspective of facility stakeholders.