Design of Satellite Attitude Control System (ACS) that involves plant uncertainties and large angle manoeuvres following a stringent pointing control, may require new non-linear control techniques in order to have adequate stability, good performance and robustness. In that context, experimental validation of new non-linear control techniques through prototypes is the way to increase confidence in the controller designed. The Space Mechanics and Control Division (DMC) of INPE is constructing a 3-D simulator to supply the conditions for implementing and testing satellite ACS hardware and software. The 3-D simulator can accommodate various satellites components; like sensors, actuators, computers and its respective interface and electronic. Depending on the manoeuvre the 3-D simulator plant can be highly non-linear and if the simulator inertia parameters are not well determined the plant also can present some kind of uncertainty. As a result, controller designed by linear control technique can have its performance and robustness degraded, therefore controllers designed by new non-linear approach must be considered. This paper presents the application of the State-Dependent Riccati Equation (SDRE) method in conjunction with Kalman filter technique to design a controller for the DMC 3-D satellite simulator. The SDRE can be considered as the non-linear counterpart of Linear Quadratic Regulator (LQR) control technique. Initially, a simple comparison between the LQR and SDRE controller is performed. After that, practical applications are presented to address problems like presence of noise in process and measurements and incomplete state information. Kalman filter is considered as state observer to address these issues. The effects of the plant non-linearities and noises (uncertainties) are considered in the performance and robustness of the controller designed by the SDRE and Kalman filter. The 3-D simulator simulink-based model has been developed to perform the simulations examples to investigate the SDRE controller performance using the states estimated by the Kalman filter. Simulations have demonstrated the validity of the proposed approach, once the SDRE controller has presented good stability margin, great performance and robustness.