1. A new methodological approach for detailed study of the organization of the cerebellar corticonuclear projection was evaluated in barbiturate-anaesthetized cats. Extracellular field potentials were simultaneously recorded in nucleus interpositus anterior and in the forelimb area of the C3 zone, at the cerebellar surface. On electrical and natural stimulation of the forelimb skin, the evoked positive field potentials in the nucleus and the climbing fibre field potentials in the cerebellar cortex had similar characteristics, indicating that the nuclear potentials were related to climbing fibre activity. 2. Application of a local anaesthetic to the cerebellar surface reversibly diminished the positive field potentials in the nucleus, demonstrating that the potentials were dependent on cerebellar cortical activity. It was thus concluded that the positive field potentials were mainly generated by climbing fibre-activated Purkinje cells and reflected synaptic inhibitory potentials in nuclear neurones. Accordingly, the positive field potentials in the nucleus could be used to reveal the termination area of Purkinje cells activated by a specific climbing fibre input evoked on peripheral stimulation. 3. The topographical organization of the cerebellar cortical projection to the forelimb part of nucleus interpositus anterior was then investigated by systematically mapping the cutaneous tactile and nociceptive 'receptive fields' of the positive field potentials at different sites in the nucleus. Five groups of receptive fields were distinguished and tentatively divided into a total of nineteen subgroups. 4. Each group of receptive fields corresponded to one or two of the previously described receptive field classes of climbing fibres to the C1, C3 and Y zones and was represented in a single area of the nucleus. Within each area there was an orderly representation of different receptive fields. The results suggest that microzones in the C1, C3 and Y zones with similar climbing fibre input project to a common set of neurones in nucleus interpositus anterior. 5. We propose a modular organization for the cerebellar control of forelimb movements through the rubrospinal tract. Each module would consist of a set of neurones in nucleus interpositus anterior and their afferent microzones in the C1, C3 and Y zones. A module would control a specific group of muscles and receive a homogeneous climbing fibre input related to the movement controlled.