It may occur in the high-speed train many different types of failures such as support fixture crack, engine fault, composite train`s division/connection abnormality, axle`s rust, and shaking of its body. Such failures can threaten safe and reliable train operation. Sometimes some failure can cause failure of the other, and therefore discovering the association rules between various failures can help preventing the occurrence of related failures. In this paper, we propose to apply association rule mining for failure record data from a high-speed train.