Received December 28, 1981 Accepted for Publication March 16, 1982 An auto-tuning method for optimizing regulator parameters of the control system in a nuclear power plant is presented. This method is based on control engineering techniques such as system identification for model estimation from measured plant response data and nonlinear optimization for optimal search of regulator parameters using a dynamic simulator of the control system with estimated models. The former technique uses a least-squares identification algorithm and the latter, a direct search simplex algorithm. A special feature of the developed tuning method is that selected performance parameters such as overshoot value, time to attain a peak, and the integral of the error squared between demand and controlled value can be incorporated into an optimization criterion. By simulated results, using actual feedwater control test data of the Boiling Water Reactor IV, the method usefulness and generality are confirmed. This auto-tuning method is found applicable not only to control system tuning in a nuclear power plant but also to that in other industrial fields.