1. Thresholds for the perception of postural sway induced by gentle perturbations were determined for five normal standing subjects. In this context we understand 'perception' to mean 'able to give a subjective report'. The thresholds for the perception of movements that were equivalent to sway in velocity and amplitude were determined when the available sensory input was limited to only one, or a pair, of the vestibular, visual, and proprioceptive systems. To examine vestibular inputs alone, vision was excluded and the whole body was moved with the ankles in a fixed position. To examine visual inputs alone, the body was kept stationary and a 'room' was moved around the subjects to simulate the relative visual-field movement that occurs during standing. To limit the available sensory input to proprioception from the legs, subjects were held stationary and balanced a load that was equivalent to their own body using their ankles. In this situation, perturbations were applied to the 'equivalent body' and these could only be perceived from the resulting ankle movements. Thresholds for perceiving ankle movements were also determined in the same posture, but with the leg muscles bearing no load. 2. The thresholds for the perception of sway during standing were very small, typically 0.003 rad at a velocity of 0.001 rad s-1, and even smaller movements were perceived as the mean velocity of the sway increased up to 0.003 rad s-1. No difference was found between the thresholds for perceiving forward sway and backward sway. Eye closure during standing did not affect the threshold for perceiving sway. 3. When sensory input was limited to proprioception from the legs, the thresholds for the perception of passive ankle movements were equivalent to the thresholds for the perception of sway during standing with all sensory inputs available. When the leg muscles were relaxed, the thresholds for perceiving ankle movements increased approximately twofold. 4. The visual thresholds for perceiving movement were higher than the proprioceptive thresholds at slower velocities of movement, but there was no difference at higher velocities. 5. Both the proprioceptive and visual thresholds were sufficiently small to allow perception of the sway that was recorded when the subjects stood normally in a relaxed manner. In contrast, the vestibular thresholds were an order of magnitude greater than the visual or proprioceptive thresholds and above the largest sway movements that were recorded during normal standing.(ABSTRACT TRUNCATED AT 400 WORDS)