The lamprey normally swims with the dorsal side up. Illumination of one eye shifts the set-point of the vestibular roll control system, however, so that the animal swims with a roll tilt towards the source of light (the dorsal light response). A tilted orientation is often maintained for up to 1 min after the stimulation. In the present study, the basis for this behaviour was investigated at the neuronal level. The middle rhombencephalic reticular nucleus (MRRN) is considered a main nucleus for the control of roll orientation in lampreys. Practically all MRRN neurons receive vestibular and visual input and project to the spinal cord. Earlier extracellular experiments had shown that optic nerve stimulation potentiates the response to vestibular stimulation in the ipsilateral MRRN. This most likely represents a neural correlate of the dorsal light response. Experiments were carried out in vitro on the isolated brainstem of the silver lamprey (Ichthyomyzon unicuspis). MRRN cells were recorded intracellularly, and the overall activity of descending systems was monitored with bilateral extracellular electrodes. The responses to 10 Hz optic nerve stimulation and 1 Hz vestibular nerve stimulation, and the influence of optic nerve stimulation on the vestibular responses, were investigated. In most preparations, optic nerve stimulation excited practically all ipsilateral MRRN cells. After stimulation, the cell was typically depolarized and showed an increased level of synaptic noise for up to 80 s. In contralateral MRRN neurons, optic nerve stimulation usually evoked hyperpolarization or no response. Vestibular nerve stimulation evoked compound excitatory postsynaptic potentials (EPSPs) or spikes in -90% of the cells, both ipsilaterally and contralaterally. A smaller subpopulation of MRRN cells (-10%) received vestibular inhibition. In 26 of 48 recorded MRRN cells, the response to vestibular stimulation was potentiated after ipsilateral optic nerve stimulation. The potentiation was seen in cells receiving either excitatory or inhibitory vestibular input as an increase in EPSP amplitudelspiking (85%) and a decrease in inhibitory postsynaptic potential amplitude (15%) respectively. In most cases the vestibular responses did not return to control levels during the testing period (10–30 min), and thus the visual stimulation most likely induced long-lasting changes in the functional connectivity of the roll control network, in addition to the short-lasting afteractivity. In four of the 11 cells recorded contralateral to the stimulated optic nerve, a depression of the vestibular response could be seen. In potentiated cells, single vestibular pulses often evoked longer episodes of large synaptic noise and sometimes spiking. In the latter case, the action potentials appeared with highly variable latency after each stimulation pulse. This indicates that an important mechanism underlying the potentiation may be a long-lasting increase in excitability in a pool of unidentified interneurons located either upstream of the MRRN cells, relaying vestibular and visual inputs, or downstream, providing positive feedback.