Event correlation analysis for alarm data at the ethylene plant in Japan was applied to reduce the number of plant alarms. This is a data-mining method that detects statistical similarities among discrete events of alarms or operations. By grouping correlated events based on the degree of similarity, countermeasures for reducing alarms can be considered more easily than by analyzing individual events of alarms and operations. By using the event correlation analysis, event data of alarms or operations of the ethylene plant was divided into a limited number of groups. The result of the analysis was helpful for estimating the effect of each countermeasure to reduce unnecessary alarms such as sequential alarms buried in the noisy plant data.