1. We compared the spatial pattern of shortest latency somatosensory (tactile) projections to the Purkinje cell (PC) layer and to the underlying granule cell (GC) layer in tactile areas of rat cerebellar cortex. Micro-mapping methods were used to sample single units in the PC layer and multiple units in the GC layer of both anesthetized and unanesthetized rats. Mechanical and electrical stimulation of the body surface were employed. Responsiveness of PCs to cutaneous stimulation was assessed by constructing histograms of simple spike activity and statistically comparing poststimulus activity to nonstimulated base-line PC activity. 2. We found that PCs respond to tactile stimulation with increases (7-10 ms) followed by decreases (8-15 ms) in simple spike activity. Increases in simple spike activity followed activation of the underlying GC layer by 1-4 ms, while decreases in simple spike activity were found 2-5 ms after GC layer activation. 3. PCs were found to have both excitatory and inhibitory receptive fields (RFs). Excitatory RFs were restricted to small areas of a single body part and for each PC were very similar or identical to the RFs of neurons in the immediately subjacent GC layer. Inhibitory PC RFs were larger, often containing more than one body part and for each PC, were only partially similar to the RFs of subjacent GCs. PC inhibitory RFs also often included body surfaces projecting to the nearby but not to the underlying GC layer. 4. Stimulation of a single peripheral locus resulted in small, distinct regions of PC layer excitation and inhibition. Areas of PC excitation overlie activated regions of the GC layer, while inhibited PCs overlie both activated and nonactivated GC regions. 5. We found PCs to be organized in groups or patches with respect to the specific body region that was capable of activating them (upper lip, lower lip, etc.). Adjacent patches of PCs often represented widely separated body parts. This pattern of PC layer activating RF projections was congruent with the pattern of excitatory RF projections to the underlying GC layer. 6. These results indicate that there is a vertical organization in GC-PC excitatory relations, while GC-induced PC inhibition is slightly more widely distributed. 7. Our finding that the patchlike activation of PCs is congruent with that of the underlying GC layer contrasts with the classical concept that PCs are activated by parallel fibers in a "beamlike" fashion from a patch of GCs. Thus, a reevaluation of the role of parallel fibers seems to us to be in order. 8. In conclusion, our results support the view that short-latency afferent tactile projections to both the GC and PC layers of cerebellar cortex are highly organized spatially. This specificity of body surface projections must be incorporated into modern views of the functional organization of cerebellar cortex.