In this study, a novel optimal anti-windup control scheme is proposed to resolve the input constraint problem for multi-input multi-output (MIMO) systems, without losing the good tracking performance as possible. Avoiding complex numerical calculation, the proposed optimal control scheme can systematically compress a huge control input within the range of the required restriction by adjusting well-designed weighted matrices for linear constrained systems. Combining a conventional linear quadratic analogue tracker (LQAT) with the proposed scheme, the anti-windup-based optimal analogue tracker not only possesses the advantage of the conventional LQAT, good tracking performance, but also effectively resolves the input saturation problem. For practical consideration, the proposed LQAT is implemented by adopting the prediction-based digital redesign technique to design an effective digital control of linear MIMO sampled-data systems under input constraints.