The experiments were designed to test two hypotheses and their corollaries: 1. That adaptation of EMG responses to support surface rotations is due to a decrease in the gain of proprioceptively triggered long-loop stretch reflexes (Nashner 1976), and that the adaptation is dependent on a normally functioning vestibular system (Nashner et al. 1982); 2. That EMG responses to rotations are generated primarily by vestibulo-spinal reflexes triggered by head accelerations (Allum and Pfaltz 1985) and comprise a coactivation of opposing leg muscles (Allum and Büdingen 1979). Adaptation with successive dorsi-flexive rotations of the support surface was investigated in the EMG responses of the ankle muscles, soleus (SOL) and tibialis anterior (TA), as well as the neck muscles, trapezius (TRAP) and splenius capitis (SPLEN CAP), both for normal subjects and for patients with bilateral peripheral vestibular deficit. Both normals and patients who first received the stimulus with their eyes open demonstrated decreasing activation at medium latency (ML), that is, with an onset at about 125 ms, and long latency (LL) responses with an onset ca 200 ms. This was the case for both ankle and neck muscles when the EMG response areas for the first 3 and second 7 of 10 trials were compared. Ankle muscle responses in the patients were diminished in area with respect to normals both with the eyes open and with the eyes closed. Ankle torque recordings from the patients were also smaller in amplitude, and these attenuated differently from normal torque responses. Functional coupling of the opposing ML and LL SOL and TA muscle responses was confirmed by the nearly coincident onset times and significantly correlated EMG response areas. At ML, ankle torque was highly correlated with TA activity when the influence of SOL was controlled. At LL, SOL activity was highly correlated with torque when the influence of TA was controlled. The delay of torque adaptation beyond the period of ML activity in normals, but not in the patients was attributed to the proportionally balanced coactivated muscle patterns producing a consistent force output and level of stability in normals. The results indicate that the adaptation in EMG response amplitudes during a sway stabilisation task is not dependent on a normally functioning vestibular system nor on visual inputs but rather appears to be due to a generalized habituation in the postural control system.(ABSTRACT TRUNCATED AT 400 WORDS)