For the safety improvement of a nuclear power plant, safety equipments are constructing additionally under new regulatory requirements. With the increase of safety equipments, it is important to maintain and improve reliability of equipments by inspection. To improve reliability of equipments, we developed a diagnostic system for steam turbine. This system consists of statistical model and physical model. Wih the statistical model, it is possible to detect abnormality from plant operation data and estimate abnormal parameters. With the physical model, it is possible to estimate failure equipment and failure casue. The physical model is based on rate of change of vibration amplitude with increase of rotational speed. Using the change of vibration, it is possible to diagnose failure cause by the plant operation data. With this diagnostic system, it is possible to improve reliability of equipments.