This paper proposes an efficient proportional fairness (PF) scheduling method using user distribution in MU-MIMO Tomlinson-Harashima precoding (THP). Although PF realizes fairness among users, heavy computational loads are experienced due to user selection, which must calculate the system capacity for all possible user combinations. To overcome this problem, the proposed approach partially allocates channel resources for semiorthogonal user selection (SUS), which causes unfairness while maintaining throughput maximization with low computational loads. This is inspired by the fact that user scheduling for throughput maximization such as SUS increases the degree of fairness under non- uniform user distributions. Hence, in the proposed method, the user distribution is measured by the spread of channel state information (CSI) power obtained from multiple antennas, and the resource allocation to PF and SUS is adaptively controlled based on this spread. Computer simulations are used to demonstrate the effectiveness of the proposed method in comparison with the traditional MU-MIMO THP with PF in terms of fairness among users and calculation costs.