The paper presents findings obtained by analyzing the construction fatality data occurred for the last 5 years using Data Mining method. Association rules between the attributes of construction fatalities causing the dead and serious injuries are described. The industrial accident data obtained from KOSHA(Korea Occupational Safety & Health Agency) were analyzed. A series of processes of data cleaning were carried out for using the data. Using Association Rule Mining, 581 association rules were retrieved using SAS E-miner. 74 association rules which are inferred as meaningful knowledge were discovered. It is expected that those rules can be useful knowledge and invaluable data for effective safety management and prevention of accident.