Objective The purpose of this study is to analyze the efficiency of neuro-mechanical compensation for kinematic variability during postural control.Signal dependent noise (SDN) in natural neuromuscular recruitment brought intrinsic instability in posture maintaining tasks.However, random drifting of human arm endpoint was substantially suppressed by neuro-mechanical compensation and reflex regulation in normal subjects, and it may become notable in pathological conditions like Parkinsons disease.Revealing the efficiency of neuro-mechanical compensation for kinematic variability may help us find out the control principals our brain adopted in position control, and maybe helpful for rehabilitation treatment planning for patients with motor deficits.Methods We use a realistic virtual arm (VA) model to address this issue through a computational approach.Stiffness properties at muscle, joint, and hand coordinates were evaluated to obtain an estimation of mechanical impedance.Interactions between muscle stiffness and SDN were simulated with simple patterns of feedfoward activation of muscles.The resultant stiffness and drifting in hand were analyzed by principal components and characterized by elliptic depiction.Results (t) The extent of random hand drifting ellipse was inversely modulated by the hand stiffness ellipse.(2) Higher levels of muscle activation caused anisotropic increases of hand stiffness mainly in the major axis direction, consequently leading to anisotropic growth of hand drifting.(3) A U-shaped ratio against total muscular energy consumption between the area of hand endpoint stiffness ellipse and that of hand endpoint drifting ellipse, therefore, suggests that it may not be efficient to counteract internal neural noise by using high levels of muscle stiffiness (activation) in multi-joint arm.Conclusion A suitable level of muscle stiffness may be optimal in maintaining a steady arm posture with minimal variation.