Abstract: A robust manipulation algorithm is proposed for multi-fingered hand in a constrained environment when uncertainties in the dynamic models and external disturbances are considered. The algorithm can guarantee the position tracking error, both velocity error and the force error between object and environment have the property of globally exponential convergence when the parametric uncertainties are considered. If the system is subject to external disturbances, it satisfies a given performance criterion of disturbance rejection. Furthermore, the robustness of the system rejecting external disturbance is improved by the robust optimization algorithm for the internal forces.