The paper proposes an approach to decentralized model reference adaptive control (MRAC) design for nonlinear systems. The proposed method is based on State Dependent Riccati Equation (SDRE) approach which is an emerging design technique for the control of nonlinear systems. A nonlinear reference model is used for the adaptive control design. The plant dynamics is assumed to be nonlinear and an adaptation methodology is designed such that response of the nonlinear reference model is followed. The adaptation is performed on a state dependent basis mainly using the adaptation mechanism designed for multi input multi output (MIMO) linear time invariant (LTI) systems. Hence, a state dependent adaptation is included to the existing adaptation mechanism of the MIMO LTI systems giving a similar design approach for the MRAC of nonlinear systems. The proposed methodology is exemplified by using the simulation model of a nonlinear system, showing the effectiveness of the SDRE based MRAC for MIMO nonlinear systems.