To detect instrument failures in a nuclear power plant, a failure detection and isolation (FDI) method based on the Kalman filter is developed. Each filter is designed to be insensitive to the failed measurements by decreasing the Kalman gain artificially. Since it is mainly dependent upon the dynamic model and averaged outputs, it can exactly indicate the direction of failures. Even though this concept minimizes the number of filters, it performs the role of analytic redundancy for estimation. As soon as the residual exceeds the predetermined bound, the Kalman filter indicates the possibility of failures. However, since the measurement may show false indication owing to abrupt noises, it must be confirmed several times by the multiple consecutive miscomparison counter, which is strongly dependent on measurement history. Then, if it is not in accordance with other measurements, detailed information on the status of the measurements is provided to help the operator's decision. Various simulations were performed to verify and validate the FDI logic in detecting steam generator and pressurizer instrument failures. It is shown that the FDI technique can detect not only a single failure but also simultaneous common-mode and sequential multiple failures of several direct redundancies. It can correctly estimate the physical states in real time, and the remaining time may be used for control with signal validation.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>