Train derailment is an important factor to influence train-running safety. In recent 10 years, derailment accounts for about 70% of railway heavy and major accidents in China. Because of the complexity of derailment and difficulty of its study, the causes of derailment are usually difficult to be found out. The traditional analytical methods are simplified, which are not able to consider the mutual influence of the factors and therefore cannot explain the essence of derailment. In this paper, a new method, intelligent synthesis is employed to analyze the essential reasons of derail accidents. Some new techniques of artificial intelligence are applied in the analysis of derailment, such as compute intelligence, fuzzy logic, etc. A basic structure of the intelligent derailment analysis support system is established, and a knowledge-ware for analyzing derailment is designed. Finally an example of actual derailment accident is analyzed by using the expert system based on the fuzzy logic method.