A dynamic predictor is proposed for linear time invariant disturbed systems with delay in the control input. The proposed dynamic predictor has a structure similar to the Luenberger observer. It outperforms the current predictors because it does not need the full state, as long as the plant is observable, it only requires the output from the system. The controller is proposed making use of the state vector from the dynamic predictor. The full control scheme including predictor and control law is applied in simulation. The advantages of the proposed predictor include robustness against external disturbances, and also the fact that the controllers developed making use of predictor are capable of tracking time-variable signals.