The interference minimization approach to pre-coding is an appropriate scheme to mitigate the interference efficiently while it may cause the capacity loss of the desired channel. On the other hand, the precoding scheme for the maximal capacity of the desired channel improves the capacity of the desired users while it increases the interference power and finally causes the capacity loss of the interfered users. Therefore, we propose a precoding scheme which satisfies these two conflicting goals and manages the interference signal in an interference-limited environment such as a multiple-input multiple-output (MIMO)-based cognitive radio (CR) network. The proposed scheme consists of two steps. The first step nulls out the largest singular value of the interference channel to mitigate the dominant interference signal based on the interference minimization approach. The second step calculates the total system capacity of each mode and selects a mode to maximize the total system capacity. In the second step, each mode consists of the right singular vectors corresponding to the singular values except the largest singular value eliminated in the first step. Simulation results show that the proposed precoding scheme not only efficiently mitigate the interference signal, but also has the best performance in terms of the total system capacity in a MIMO-based CR network.