Interactions between proprioceptive and vestibular inputs contributing to the generation of balance corrections may vary across muscles depending on the availability of sensory information at centres initiating and modulating muscle synergies, and the efficacy with which the muscle action can prevent a fall. Information which is not available from one sensory system may be obtained by switching to another. Alternatively, interactions between sensory systems and the muscle to which this interaction is targeted may be fixed during neural development and not switchable. To investigate these different concepts, balance corrections with three different sets of proprioceptive trigger signals were examined under eyes-open and eyes-closed conditions in the muscles of normal subjects and compared with those of subjects with bilateral peripheral vestibular loss. The different sets of early proprioceptive inputs were obtained by employing three combinations of support surface rotation and translation, for which ankle inputs were nulled, normal or enhanced, the knees were either locked or in flexion, and the trunk was either in flexion or extension. Three types of proprioceptive and vestibulospinal interactions were identified in muscles responses. These interactions were typified by the responses of triceps surae, quadriceps, and paraspinal muscles. The amplitudes of stretch responses at 50 ms after the onset of ankle flexion in triceps surae muscles were related to the velocity of ankle stretch. The amplitude of balance-correcting responses at 100 ms corresponded more with stretch of the biarticular gastrocnemius when the knee was re-extended at 60 ms. Absent stretch reflexes at 50 ms in triceps surae with nulled ankle inputs caused a minor, 12-ms delay in the onset of balance-correcting responses in triceps surae muscles. Vestibular loss caused no change in the amplitude of balance-correcting responses, but a negligible decrease in onset latency in triceps surae even with nulled ankle inputs. Stretch responses in quadriceps at 80 ms increased with the velocity of knee flexion but were overall lower in amplitude in vestibular loss subjects. Balance-correcting responses in quadriceps had amplitudes which were related to the directions of initial trunk movements, were still present when knee inputs were negligible and were also altered after vestibular loss. Stretch and unloading responses in paraspinals at 80 ms were consistent with the direction of initial trunk flexion and extension. Subsequent balance-correcting responses in paraspinals were delayed 20 ms in onset and altered in amplitude by vestibular loss. The changes in the amplitudes of ankle (tibialis anterior), knee (quadriceps) and trunk (paraspinal) muscle responses with vestibular loss affected the amplitudes and timing of trunk angular velocities, requiring increased stabilizing tibialis anterior, paraspinal and trapezius responses post 240 ms as these subjects attempted to remain upright. The results suggest that trunk inputs provide an ideal candidate for triggering balance corrections as these would still be present when vestibular, ankle and knee inputs are absent. The disparity between the amplitudes of stretch reflex and automatic balance-correcting responses in triceps surae and the insignificant alteration in the timing of balance-correcting responses in these muscles with nulled ankle inputs indicates that ankle inputs do not trigger balance corrections. Furthermore, modulation of balance corrections normally performed by vestibular inputs in some but not all muscles is not achieved by switching to another sensory system on vestibular loss. We postulate that a confluence of trunk and upper-leg proprioceptive input establishes the basic timing of automatic, triggered balance corrections which is then preferentially weighted by vestibular modulation in muscles that prevent falling. (ABSTRACT TRUNCATED)