In this article, a solution is presented for the input decoupling problem with a control input consisting of a proportional-plus-derivative (PD) function of the signal to be rejected, with the requirement of internal stability of the closed-loop, in the general case where all the feedthrough matrices are possibly non-zero. This problem coincides with the so-called decoupling with preview in the discrete-time case. However, in this article the continuous and discrete cases are treated in a unified setting. A simple synthesis technique is also presented for the computation of the gain matrices appearing in the decoupling control law. This technique does not require the extra assumption of left invertibility, or changes of coordinates. In the last part of this article, the implementation of the decoupling control is discussed both via mixed feedback/feedforward and pure feedforward control architectures.