This paper presents an iterative post-processing scheme for conventional K-best MIMO detection based on repeating the K-best search from a single root node. Utilizing the partial Euclidean distance and the detected path in the conventional detection process, the root node can be determined for each post-processing iteration. Once the correct root node is determined in post-processing, we can use a smaller value of K in the post-processing process to obtain the correct path. Moreover, the proposed scheme is only enabled when the partial Euclidean distance of the detected path is extremely large and considered to be incorrect to reduce the latency and the computational complexity. Based on simulation results, the proposed method provides a notable improvement in performance while maintaining a similar complexity for MIMO systems using high-order modulations.