Correspondence analysis (CA) is generic statistical tool for mathematical as well as visual analysis of categorical data. In this study, its usefulness is demonstrated by analyzing steel plant
derailments data. The considerations in the data structure and
selection of models can be applied to any generic plant for derail-ment data analysis. This study brings about previously unexplored issues with the plant specific rail transportation system and vali-dates the reasons for derailments. The uniqueness of this study is identifying four meaningful rules in explaining the root causes of
derailments. Rule 1 can be implemented by optimally balancing the rail movements, i.e., scheduling the rail logistics carefully. To reduce the human involvement in derailments incidents (rule 2),
mechanization is recommended. Increasing production without adequate consideration of in-plant rail infrastructure should be stopped. To manage rule 3 related incidents, a common collabora-tive maintenance system using risk based maintenance philosophy is suggested. Risk based maintenance programmes are also needed to avoid rule 4 related incidents