1. In the guinea-pig, a unilateral labyrinthectomy induces postural disturbances and an ocular nystagmus which abate or disappear over time. These behavioural changes are accompanied by an initial collapse and a subsequent restoration of the spontaneous activity in the neurones of the ipsilateral vestibular nuclei. Recently, it has been shown that the vestibular neuronal activity remained collapsed over at least 10 h whereas its restoration was complete 1 week after the lesion. The aims of this study were to determine when restoration of spontaneous activity in the partially deafferented vestibular neurones started and to compare the time courses of the behavioural and neuronal recoveries in guinea-pigs that had undergone a unilateral labyrinthectomy. 2. Neuronal discharge measurements were made using chronic extracellular recording of single unit activity. After a left labyrinthectomy, electrodes, were placed on the site of the destroyed labyrinth to enable stimulation of the left vestibular nerve. Behavioural measurements included chronic recording of eye movements by the scleral search coli technique. After a left labyrinthectomy, lateral deviation of the head, twisting of the head, and eye velocity of the slow phases of the nystagmus were measured. 3. The neuronal activity of the rostral part of the vestibular nuclear complex on the lesioned side was recorded in alert guinea-pigs over 4 h recording sessions between 12 and 72 h after the lesion. 4. The criterion used to select vestibular neurones for analysis was their recruitment by an electric shock on the vestibular nerve. In addition, in order to explore a uniform population, we focused on neurones recruited at monosynaptic latencies (0.85-1.15 ms). 5. For each recording period, the mean resting rate was calculated animal by animal and the grand mean of these individual resting rate means was calculated. Previously, a decline in the grand mean resting rate from 35.8 +/- 6.0 spikes s-1 (control state) to 7.1 +/- 4.2 spikes s-1 during the first 4 h after labyrinthectomy has been shown. In the present study, the first sign of recovery was observed during the 12-16 h recording period when the resting rate grand mean increased to 16.3 +/- 3.9 spikes s-1. This grand mean activity did not change significantly during the following 12 h. Thereafter, restoration of neuronal activity improved and was complete 1 week after the lesion. 6. Although the abatement of the vestibular symptoms roughly paralleled the restoration of neuronal activity in the vestibular nuclei, some discrepancies between the time courses of both phenomena emerged. An important step in postural recovery (the animals managed to stand up) and a major part of the abatement of the nystagmus occurred before the recovery of vestibular neuronal activity. In addition, lateral deviation of the head disappeared while restoration of the neuronal activity was incomplete, but significant head twisting was still evident when vestibular resting rates had recovered completely. 7. We conclude that restoration of neuronal activity in the ipsilateral vestibular nuclei starts 12 h after the lesion and that restoration of neuronal activity in the ipsilateral vestibular nuclei is not the only mechanism underlying behavioural vestibular compensation.