When the doubly-fed induction generators (DFIGs) operate under unbalanced and distorted power grid conditions, the traditional control strategy can hardly maintain its original stable performance, resulting in distortion of the stator and rotor currents. In addition, when the motor parameters are affected by environmental factors and the machine's own design limitations, the performance of the controller used will also be further reduced. This paper proposes a robust predictive rotor current control (R-PRCC) scheme for a DFIG in order to obtain sinusoidal and balanced rotor current under complex operating conditions. The reference value of the positive sequence component of the rotor current is obtained by cascaded delayed signal cancellation (CDSC), and the total system disturbance caused by model uncertainty can be quickly estimated by the Extended State Observer (ESO). By combining with deadbeat control, the rotor current can be fast and accurately controlled. The proposed R-PRCC scheme is compared to conventional model predictive rotor current control (MPRCC). Simulated and experimental results obtained for a 1.5 kW laboratory DFIG system confirm the effectiveness of the proposed method.