The contribution of visual and somatosensory input to stabilization after sudden postural disturbances was investigated. Fast transient platform movements (50 degrees/sec, 4 degrees), rotating toe-up around the ankle-joint, evoke EMG responses in the triceps surae (TS) and anterior tibial (TA) muscles. The short and medium latency responses (SL, ML) in the TS destabilize, the long latency responses (LL) in the TA stabilize the upright posture. The influence of vision was tested in normals comparing the conditions eyes open with eyes closed. The influence of additional somatosensory input was tested with the operator's index finger touching the back of the subject with eyes closed. Absent visual input (eyes closed) resulted in a decrease in latency of the stabilizing LL response and an increase in the integrated EMG (iEMG). With eyes closed the latency of the LL response decreased by 4 msec on average (mean eyes open: 111.6 msec (S.D. = 17.3), eyes closed: 107.6 msec (S.D. = 17.5); P < 0.001 (Wilcoxon test)), the iEMG increased by 10% (mean eyes open: 13.9 microV.sec (S.D. = 7.6), eyes closed: 15.5 microV.sec (S.D. = 8.6); P < 0.001). Additional somatosensory input as an external reference for body orientation in space resulted in an increase in latency and decrease in iEMG: the latency increased by 7 msec on average (mean eyes closed: 107.6 msec (S.D. = 17.5), index finger and eyes closed: 114.7 msec (S.D. = 19.9); P < 0.001), the iEMG decreased by 20% (mean eyes closed 15.5 microV.sec (S.D. = 8.6), index finger and eyes closed: 12.3 microV.sec (S.D. = 7.5); P < 0.001). There was no significant change in latency of SL and ML. We conclude that postural EMG responses in leg muscles after fast transient platform movements change according to different functional demands, modulated via visual and somatosensory input.