Objective Stability of limb posture is important for motor behaviors in the presence of perturbations.Stiffness is usually used to characterize the stability of our limb posture.Reflex pathways significantly contribute to muscle stiffness.It is one of the three individual stiffness components: intrinsic, passive and reflex.Previous experiments have demonstrated that the stiffness generated by reflex action can be modulated according to the nature of motor tasks and the environments.In either case, the goal of stiffness modulation is to reduce instability by increase the stiffness.However, it has remained a technical challenge to quantify the reflex action compensates in experiments for the end-point stability of the upper limb.The present study aims to address this problem.Methods In the present study, a close-loop virtual arm model with reflex pathway was developed based on an open-loop model.Simulation with these two models allows us to separate the influence of reflex action completely from the intrinsic components.To simulate the realistic kinematic variability in human motor behaviors, dynamic simulation of these two 6-muscle, 2-DOF models were executed with signal dependent noise.The method used to evaluate stiffness is by adding perturbations to the muscle force.Using the measured changes and added forces, stiffness can be obtained by a Least-Squares method in MATLAB.Distribution of drifting in the end-point of the arm is characterized by an ellipse, whose major and minor axis are the eigen-values of covariance matrix of the end-point position.Fifteen trails of simulation experiments with different activation level patterns have been done using open-loop and close-loop models separately.All these activation patterns keep the arm posture stable in an identical position.Results By analyzing the results, we found that, by average, stiffness with reflex are larger than that without reflex by 18%.This value is agreed with the experimental data obtained previously.Also, the distribution of drifting at the end-point is reduced by 35% with reflex compared to that without reflex.The shape and orientation of major axis of the stiffness ellipse are the same no matter with or without reflex action.Conclusion Based on simulation experiments, we conclude that the reflex action can increase the stiffness and decrease kinematic variations at the end-point simultaneously.Activation pattern determines the main feature of the stiffness and the reflex action only plays an assistant role in the modulation of stiffness.