This study examines the influence of the cerebellum on the excitability of inferior olivary neurons in the cat. Two major pathways from the cerebellar nuclei to the inferior olive have been investigated by electrophysiological and anatomical techniques. The first, excitatory pathway connects the cerebellar nuclei through nuclei at the mesodiencephalic junction with the inferior olive. The second is the direct, GABAergic, nucleo-olivary pathway. Intra- as well as extracellular recordings obtained in the rostral part of the medial accessory and principal olives revealed that electrical stimulation with a short burst of three pulses delivered at the mesodiencephalic junction results in short-latency activation (4–8 ms) of most olivary neurons. More than half of the units showed, in addition to the short-latency activation, a consistent response with a much longer latency (-180 ms). Many units (66%) that responded to mesodiencephalic stimulation could also be activated by superior cerebellar peduncle stimulation with a similar stimulation paradigm (latency 9–15 ms). However, in such cases consistent long latency responses were only rarely recorded (7%). To distinguish between the effect of the two pathways, both of which are activated by superior cerebellar peduncle stimulation, an electrolytic lesion of the nucleo-olivary fibres was made in the brainstem in six experiments. The effect of this lesion was verified in three cases by retrograde horseradish peroxidase tracing from the rostral inferior olive at the end of the experiment. This time only extracellular recordings were made. Stimulation of the mesodiencephalic junction still resulted in easily activated olivary units which showed an increased probability of firing a long-latency action potential. Stimulation of the superior cerebellar peduncle now resulted in a 50% decrease in probability of activating olivary units in the short-latency range. However, a five-fold increase in the chance of triggering action potentials in the long latency interval was noted, implying that many units reacted only with a long-latency action potential. The results obtained with our experimental paradigm appear enigmatic since it is well established that the nucleo-olivary pathway is GABAergic and thus, by convention, should be inhibitory to the olivary neurons. However, it is possible to explain these results in terms of dynamic coupling of olivary neurons. This concept ascribes an important role to the nucleo-olivary pathway in regulating the degree of electronic coupling between olivary neurons (probably by a shunting mechanism) and as such may be an important instrument in the regulation of synchronous and rhythmic olivary discharges. Thus, lesion of this pathway would be expected to result in coupling of large aggregates of olivary cells. It seems likely that these strongly coupled cell ensembles are more difficult to activate by incoming afferent volleys. However, once activated, the coupled olivary neurons develop an oscillation of the membrane potential which may be conveyed, electronically, to neighbouring neurons and subsequently, during the depolarizing phase of the oscillation, result in a more easily triggered rebound or longlatency response. It is concluded that cerebellar output may not merely inhibit olivary neurons, but also, in conjunction with an excitatory nucleo-mesodiencephalo-olivary circuit, modulate olivary excitability in a rather complex manner.