Use of a gas turbine engine as the primary power source has been popular in light and heavy industries, aerospace engineering, marine engineering, etc. Gas turbine engine is also used in our modern national defense weapons in Taiwan. For instance, most of Air Force fighters use gas turbine engines as the source of power. Gas turbine engines are usually associated with various sensors for real-time condition surveillance and require periodical maintenance for providing proper functioning and safety guarantee. In contrast, real-time failure prediction of gas turbine engine components could be achieved by applying data mining or statistics techniques. However, such failure prediction will not be effective when applying to engines which are deteriorated by long-term running in high temperature and high stress environment. In this study, we collected maintenance and operating logs according to the engine deterioration history and established and empirically evaluated four different data-mining-based prediction models. The proposed data-mining-based prediction approach attempts to predict the time-to-deterioration for a gas turbine engine after its prior deterioration occurrence, to provide maintenance personnel accurate prediction for better making or revising maintenance schedules, and to achieve the ¡§foreseeing maintenance and management policy.¡¨