This study aimed to examine the interactions of visual, vestibular, proprioceptive, and tactile sensory manipulations and sitting on either a stable or an unstable surface on mediolateral (ML) trunk sway. Fifteen individuals were measured. In each trial, subjects sat as quiet as possible, on a stable or unstable surface, with or without each of four sensory manipulations: visual (eyes open/closed), vestibular (left and right galvanic vestibular stimulation alternating at 0.25 Hz), proprioceptive (left and right paraspinal muscle vibration alternating at 0.25 Hz), and tactile (minimal finger contact with object moving in the frontal plane at 0.25 Hz). The root mean square (RMS) and the power at 0.25 Hz (P25) of the ML trunk acceleration were the dependent variables. The latter was analyzed only for the rhythmic sensory manipulations and the reference condition. RMS was always significantly larger on the unstable than the stable surface. Closing the eyes caused a significant increase in RMS, more so on the unstable surface. Vestibular stimulation significantly increased RMS and P25 and more so on the unstable surface. Main effects of the proprioceptive manipulation were significant, but the interactions with surface condition were not. Finally, also tactile manipulation increased RMS and P25, but did not interact with surface condition. Sensory information in feedback control of trunk posture appears to be reweighted depending on stability of the environment. The absolute effects of visual and vestibular manipulations increase on an unstable surface, suggesting a relative decrease in the weights of proprioceptive and tactile information.