The active fault-tolerant control under random loss of measurement data issue is investigated for a class of networked control systems with actuator faults and partially decoupled disturbances. First, the equivalent state-augmented systems are derived using an appropriate model transformation, and an unknown input observer (UIO) is designed to simultaneously estimate system states and fault signals under random packet dropouts. Next, considering the online estimation of states and fault, an active fault-tolerant control law is developed using a signal compensation strategy. Intensive stability analysis is performed to obtain sufficient conditions to further generate the desired observer and controller. Furthermore, the corresponding optimal parameters can be co-designed by solving the matrix inequality with convex constraints online. Finally, a simulation example of the jet engine model is provided to demonstrate the effectiveness of the proposed fault estimation and fault-tolerant control method.