In this paper a novel model predictive control (MPC) approach is proposed based on mixed-H 2 ∕H ∞ control for space teleoperation systems with unknown, large time-varying delays and input constraints. This novel approach only measures the online delay; it does not presume the delay bounds. In addition, H ∞ control has been incorporated into the original MPC, which compensates for large timevarying delays, handles input constraints, and provides the desired tracking performance. First, the space teleoperation system is built based on a type of control architecture and then converted into a discrete state-space equation. Next, a state-feedback controller is designed based on linear matrix inequality (LMI). Meanwhile, the corresponding sufficient conditions are derived for the controller, enabling the closed-loop system to be asymptotically stable and guaranteeing the prescribed mixed-H 2 ∕H ∞ performance under input constraints. The parameters of the controller are calculated online by updating the mixed-H 2 ∕H ∞ optimizing problem during each interval. Finally, comparative simulations reveal that the proposed approach can achieve better performance in terms of tracking ability than that of traditional MPC. Moreover, it can ensure the input constraints in the unknown, large time-varying delay scenario whereas the original H ∞ control approach cannot guarantee performance.