© IFAC.Model predictive control (MPC) is investigated as a control method for frequency control of power systems which are exposed to increasing wind power penetration. For such power systems, the unpredicted power imbalance can be assumed to be dominated by the fluctuations in produced wind power. An MPC is designed for controlling the frequency of wind-penetrated power systems, which uses the knowledge of the estimated worst-case power imbalance to make the MPC more robust. This is done by considering three different disturbances in the MPC: one towards the positive worst-case, one towards the negative worst-case, and one neutral in the middle. The robustified MPC is designed so that it finds an input which makes sure that the constraints of the system are fulfilled in case of all three disturbances. Through simulations on a network with concentrated wind power, it is shown that in certain cases where the state-of-the-art frequency control (PI control) and nominal MPC violate the system constraints, the robustified MPC fulfills them due to the inclusion of the worst-case estimates of the power imbalance.