As with any alerting system, for infusion pump alerts there is a tradeoff between the number of false alarms and the number of correct detections produced by the system. That is, if we try to eliminate false alarms, the result may be more missed detections; if we try to eliminate missed detections, the cost may be more false alarms. Since missed detections are very costly in the healthcare system, the tendency is to allow more false alarms while driving missed detections to zero. However, performance problems may result from too many false alarms. I will present the results of research from the aviation domain that discusses the effect of false alarms on performance, showing that there are certain conditions that tend to produce errors in responding to alerts. In addition, we are investigating the application of a method where alerting thresholds are set to achieve an acceptable or even optimal balance of false alarms and correct detections. I will discuss how this method is being applied and what guidance we expect it to produce.