Since suitable control of water level can greatly enhance the operation of a power station, a Fuzzy logic controller architecture is applied to show desired control of the water level in a Nuclear steam generator. With regard to the physics of the system, it is shown that two inputs, a single output and the least number of rules ( 9 rules ) are considered for a controller, and the ANFIS training method is employed to model functions in a controlled system. By using ANFIS training method, initial member functions will be trained and appropriate functions are generated to control water level inside the steam generators while using the stated rules. The proposed architecture can construct an input – output mapping based on both human knowledge (in from of Fuzzy if – then rules) and stipulated input – output data. In this paper with a simple test it has been shown that the architecture fuzzy logic controller has a reasonable response to one step input at a constant power. Through computer simulation, it is found that Fuzzy logic controller is suitable, especially for the water level deviation and abrupt steam flow disturbances that are typical in the existing power plant. /3/, /6/