In this paper, the stabilization problem of loaded double inverted pendulum using an optimized state feedback sliding mode control is investigated. ADAMS/Matlab co-simulation environment is used for building a virtual nonlinear model for loaded double inverted pendulum system and the state feedback sliding mode control law is designed for stabilizing the system. Mismatched uncertainties represented by payload variations is considered. Genetic Algorithms are used to optimize the parameters of the sliding mode controller based on a performance index containing the sum of squared errors. The proposed control scheme can significantly suppress chattering effect and improves the performance of the system against uncertainties. Simulation results show the effectiveness of the approach and the robustness of the system against payload changes.