1. Neuronal activity was investigated in the left superior vestibular nucleus (SVN), lateral vestibular nucleus (LVN), and rostral part of the medial vestibular nucleus (MVN) in the alert guinea pig after a unilateral (left) labyrinthectomy was performed. Vestibular neurons were recorded either immediately (just-postoperative group, n = 6) or 1 wk after labyrinthectomy (1-wk-postoperative group, n = 6) and compared with the activity recorded in intact animals (control group, n = 6). 2. Animals were prepared for extracellular recording of single-unit activity and for eye movement recording (scleral search coil technique). To enable stimulation of the left vestibular nerve, bipolar silver ball electrodes were chronically implanted either in contact with the bony labyrinth in the control group or close to the stump of the vestibular nerve after labyrinthectomy. Complete labyrinthectomy was performed under halothane anesthesia. 3. The criterion used to select vestibular neurons for analysis was their recruitment by an electric shock on the vestibular nerve. Of the 589 recorded neurons, 424, defined as second-order vestibular neurons, were recruited at monosynaptic latencies (0.85-1.15 ms) and 165 were recruited at polysynaptic latencies. One hundred three second-order vestibular neurons were recorded in the control group, 173 in the just-postoperative group, and 148 in the 1-wk-postoperative group. 4. The activity of the electrically recruited neurons was recorded during sinusoidal horizontal head rotation in the dark (0.3 Hz, 40 degrees/s peak velocity). The behavior of the neurons was analyzed by plotting their firing rate against head velocity. The Y-intercept of the regression line was used to express spontaneous firing rate (resting discharge), and its slope was used to express the sensitivity of the neuron-to-head velocity. 5. In the absence of statistically significant difference between the characteristics of the neuronal discharge of the second-order vestibular neurons recorded in the SVN, LVN, and rostral MVN, the data were pooled. The Resting discharge of these cells amounted to 41.0 +/- 24.7 (SD) spikes/s in the control state, fell to 7.2 +/- 13.9 spikes/s just after labyrinthectomy, and completely returned to normal values 1 wk after surgery (42.5 +/- 21.6 spikes/s). Among the monosynaptically recruited neurons, the percentage of silent units was 0% in the control group, 69% in the just-postoperative group, and 0% in the 1-wk-postoperative group. 6. By contrast, the sensitivity to head velocity of the second-order vestibular neurons, which was 0.69 +/- 0.48 (SD) spikes.s-1/deg.s-1 in the control state and which fell to 0.03 +/- 0.11 spikes.s-1/deg.s-1 just after labyrinthectomy, remained low 1 wk after injury (0.21 +/- 0.26 spikes.s-1/deg.s-1). Moreover, the slight recovery of sensitivity to head rotation was due only to units behaving as type II neurons. 7. The mean resting discharge of the polysynaptically recruited neurons (pooled from the 3 explored nuclei) was 31.6 +/- 19.3 spikes/s in the control group. It decreased to 11.6 +/- 12.1 spikes/s in the just-postoperative group and recovered to 39.8 +/- 20.2 spikes/s in the 1-wk-postoperative group. No neuron was silent at rest either in the control group or in the 1-wk-postoperative group. Just after labyrinthectomy, 35% of the neurons had a null resting activity. The mean sensitivity to head velocity of these neurons was 0.55 +/- 0.42 spikes.s-1/deg.s-1 in the control group. It decreased to 0.05 +/- 0.12 spikes.s-1/deg.s-1 in the just-postoperative group and recovered to 0.22 +/- 0.17 spikes.s-1/deg.s-1 in the 1-wk-postoperative group. 8. We conclude that, at least in the guinea pig, the restoration of the spontaneous activity of the deafferented neurons is complete 1 wk after a unilateral labyrinthectomy and thus probably plays an important role in vestibular compensation...