With the continuous advancement of technology, the complexity of key areas such as power systems and industrial control systems is also increasing. The real system generally faces the problems of time delay and disturbance in the actual operation process. For systems with input delays and unknown disturbances, we first analyze the existing state prediction schemes. All existing prediction schemes assume that the system states are measurable, but most system states are difficult to obtain in reality. Therefore, before proposing the new prediction scheme, a state observer that can accurately obtain the current state of the system is proposed to prepare for the subsequent state prediction. After that, this paper proposes a new state prediction scheme, which can provide different conditions to be satisfied according to the user’s demand for accuracy, and has high flexibility. Finally, a numerical example is given to demonstrate the effectiveness of the proposed state observer and prediction scheme.