1. The inhibitory effects of the GABAA agonist muscimol and the GABAB agonist baclofen on tonically active medial vestibular nucleus (MVN) neurones were recorded in slices of the rat dorsal brainstem in vitro, to determine whether any changes occurred in the functional efficacy of GABAergic inhibition in these cells during the initial rapid stage of 'vestibular compensation', the behavioural recovery that takes place after unilateral labyrinthectomy (UL). These experiments were carried out in preparations where the midline was cut, severing all commissural connections between the two vestibular nuclei. 2. Slices of the MVN were prepared from normal animals and animals that had been unilaterally labyrinthectomised 4 h earlier. The mean in vitro discharge rate of MVN neurones in the rostral region of the ipsi-lesional nucleus after UL was significantly higher than that in control slices, confirming our earlier reports of an increase in intrinsic excitability of these cells in the early stage of vestibular compensation. The in vitro discharge rates of caudal ipsi-lesional MVN cells, and rostral and caudal contra-lesional MVN cells, were not different from controls. 3. Muscimol and baclofen caused reversible, dose-related inhibition of the tonic discharge rate of MVN cells in control slices. In slices prepared from UL animals, MVN cells in the rostral region of the ipsi-lesional nucleus showed a marked downregulation of their response to both muscimol and baclofen, seen as a rightward shift and a decrease in slope of the dose-response relationships for the two agonists. In the contra-lesional nucleus, there was a small but significant upregulation of the responsiveness of both rostral and caudal MVN cells to baclofen, and a marked upregulation of the responsiveness of caudal MVN cells to muscimol. 4. In slices from animals that had undergone bilateral labyrinthectomy 4 h earlier, the downregulation of the functional efficacy of GABA receptors in the rostral MVN cells did not occur. The changes in GABA receptor efficacy after UL are therefore not due to the vestibular de-afferentation itself, but are instead due to the imbalance in excitability of the vestibular nuclei of the lesioned and intact sides, and the enhanced commissural inhibition of the ipsi-lesional MVN cells that follows UL. 5. The downregulation of GABA receptor efficacy in the ipsi-lesional MVN neurones is functionally compensatory, in that their response to commissural and cerebellar inhibitory drive will be significantly reduced after UL. Their intrinsic membrane conductances, and their remaining excitatory synaptic inputs, will consequently be more effective in causing depolarisation and the restoration of resting activity. Simultaneously the upregulation of GABAergic efficacy in the contra-lesional MVN will tend to reduce the hyperactivity on the contralateral side. These adaptive changes therefore represent a plausible cellular mechanism for the recovery of resting discharge in the ipsi-lesional MVN neurones, and the 're-balancing' of the excitability of the vestibular neurones of the lesioned and intact sides, as occurs after UL in vivo. 6. We propose that the adaptive regulation of the functional efficacy of GABA receptors in the MVN neurones may be an important cellular mechanism for the 'homeostasis of bilateral excitability' of the vestibular nuclei of the two sides.