Although balance training can improve balance across various populations, the underlying mechanisms, such as how balance training may alter sensory integration, remain unclear. This study examined the effects of balance training with visual input manipulations provided by virtual reality versus conventional balance training on measures of postural sway and sensory integration during balance control. Twenty-two healthy young adults were randomly allocated into a balance training group (BT) or a balance training with virtual reality group (BT + VR). The BT received traditional balance training, while the BT + VR additionally received visual manipulations during the 4-week balance training to elicit sensory conflicts. Static balance was measured in the form of center of pressure (COP) sway speed in trained (eyes open) and untrained (eyes closed) balance conditions. A model-based analysis quantified the sensory integration and feedback characteristics of the balance control mechanism. Herein, the visual weight quantifies the contribution of visual orientation information to balance while the proportional and derivative feedback loop-gains correct for deviations from the desired angular position and angular velocity, respectively. Significant main time effects were observed for the visual sensory contribution to balance (p = 0.002, $$\:{\eta\:}_{p}^{2}$$ = 0.41) and for the derivative feedback loop-gain (p = 0.011, $$\:{\eta\:}_{p}^{2}$$ = 0.29). Significant group-by-time interactions were observed for COP sway speed in the untrained task (p = 0.023, $$\:{\eta\:}_{p}^{2}$$ = 0.23) in favor of BT + VR and in the proportional feedback loop-gain, with reductions only in the BT + VR group (p = 0.043, $$\:{\eta\:}_{p}^{2}$$ = 0.2). BT + VR resulted in larger performance improvements compared with traditional BT in untrained tasks, most likely due to reduced reliance on visual information. This suggests that the systematic modulation of sensory inputs leads to enhanced capacity for motor adaptation in balance training.
