In this paper, a robust sensor and actuator fault detection and isolation (FDI) method based on unknown input observers (UIO) is adopted and applied to the load frequency control loops of interconnected power systems. Changes in load demand are regarded as the unknown disturbances in the system and the designed UIOs and thus the proposed method are robust to these disturbances. Using selected different sets of measured variables in the UIO design, simulations are performed for the dynamical model of a power control system composed of two areas. As we distinguish the cases for successful and unsuccessful sensor and controller FDI, it is shown that the proposed scheme is able to detect and isolate sensor and controller faults for proper selections of measured variables. By using residuals generated by the UIOs, the designed "fault detection and isolation logic" system shows the operator which sensor or controller is faulty. Hence the faulty sensor or controller can be replaced by a healthy one for a more reliable operation.