A human disruption model is used to develop a disruption manager that mediates instant message interruptions as people navigate the Web. The system is designed to reduce disruption and increase overall user satisfaction. The disruption manager is based on research investigating people�s reactions to interruptions and factors involved in the interruption process, such as interruption relevancy and task priority. The manager supports monitoring ongoing behaviors using implicit metrics to control possible disruptive outcomes given the user and system state (mouse and keyboard behaviors, interruption type, task request or notification, concepts surrounding the user�s goals, interruption relevancy, and concept priority). The disruption manager selects appropriate interruption timing and selects whether or not instant message interruptions should be presented to the user.