The computational study of human balance recovery strategy is crucial for revealing effective strategy in human balance rehabilitation and humanoid robot balance control. In this context, many efforts have been made to improve the ability of quiet standing human balance. There are three main strategies for human balance including (i) ankle, (ii) hip, and (iii) stepping strategies. Besides, arm usage was considered for balance control of human walking. However, there exist few works about effectiveness assessment of arm strategy for quiet standing balance recovery. In this paper, we proposed a nonlinear model predictive control (NMPC) for human balance control on a simplified model with sagittal arm rotation. Three case studies including (i) active arms, (ii) passive arms, and (iii) fixed arms were considered to discuss the effectiveness of arm usage for human balance recovery during quiet standing. Besides, the total root mean square (RMS) deviation of joint angles was computed as an index of human motion intensity quantification. The proposed solution has been implemented for a human-like balance recovery with arm usages during quiet standing under perturbation and shows the effectiveness of arm strategy.