Event correlation is becoming one of the most central techniques in managing the high volume of event messages. Practically, no network management system can ignore network surveillance and control procedures which are based on event correlation. The majority of existing network management systems use relatively simple ad hoc additions to their software to perform alarm correlation. In these systems, alarm correlation is handled as an aggregation procedure over sets of alarms exhibiting similar attributes. In recent years, several more sophisticated alarm correlation models have been proposed. In this paper, we will expand our knowledge-based event correlation model to capture temporal constraints.