In this paper, we propose a new robust receding horizon control scheme for linear input-constrained discrete-time systems with polytopic uncertainty. We provide a rigorous proof for closed-loop stability. The control scheme is based on the minimization of the worst-case cost with time-varying terminal weighting matrices, which can easily be implemented by using linear matrix inequality optimization. We discuss modifications of the proposed scheme that improves feasibility or on-line computation time. We compare the proposed schemes with existing results through simulation examples.