Snider 's pioneering studies of tactile responses in the cerebellar cortex of cats and monkeys suggested that posterior regions of the cerebellar hemispheres receive somatotopically organized projections. However, recent studies in rats, using high-density, in depth microelectrode mapping methods, have shown that tactile projections to the granule cell layer of the cerebellar hemispheres are somatotopically disrupted. We reexamined the organization of cutaneous projections to cerebellar hemispheric cortex in cats by using micromapping methods. Natural stimulation of cutaneous surfaces evokes short-latency (mossy-fiber-induced) multiple unit responses in the cerebellar granule cell layer of crus II and paramedian lobule in both ketamine- and barbiturate-anesthetized cats. Facial structures are represented in several of the most caudomedial folia of crus II as well as in three of the rostral folia of the paramedian lobule. In several of these paramedian folia, facial projections are interspersed with projections from the forelimb. Forelimb structures alone are represented in two intermediate folia of the paramedian lobule. No cutaneous projections were found from the trunk or hindlimb. All projections were from ipsilateral receptive fields. In four folia of crus II and six folia of the paramedian lobule, cutaneous projections form a mosaic of patchlike projections. Within single patches, projections are somatotopically organized, but projections to adjacent patches come from noncontiguous body regions. Within a single folium , a particular facial region may be represented in two or three spatially separated patches. Facial patches are small, usually less than 1 mm2. Forelimb patches are usually larger, often extending the full length and breadth of a folial crown. Patches with like receptive fields are not organized in zonal sagittal strips. Rather than being somatotopically organized, cutaneous mossy fiber projections to granule cells in cats, as in rats, reveal a more complex mosaic pattern of organization.