Cortical afferents transferring information destined for the cerebellum terminate in the pontine nuclei (PN) in a divergent and patchy fashion. We investigated whether the form of dendritic fields of pontine projection neurons which are postsynaptic to the cortical afferents are related to this patchy pattern. To this end we used a triple combination of (1) retrograde labeling (injection of Fluorogold into the brachium pontis), (2) anterograde labeling [injection of Dil into cortical areas A17 and Sml(forelimb)], and (3) subsequent intracellular fills of identified projection neurons (Lucifer yellow) in slightly fixed slices of pontine brainstem. In 64 projection neurons whose somata were located within 160 microns of the border defined by cortical afferent fields, most of the dendritic trees were found to respect the border. Strikingly, proximal dendrites which were oriented toward the border often bent in order to avoid the boundary. This observation was supported by a quantitative analysis. It revealed that overlap areas of dendritic fields with the neighboring compartment were significantly smaller than those of hypothetical, radially organized dendritic fields of the same size, indicating that the dendritic fields are indeed confined to single compartments. In a second series of experiments, double injections of the anterograde tracers Dil and DiAsp into adjacent sites within one cortical area (A17 or Sml) were made in order to test if the topology of the cortical map is preserved within individual pontine compartments. This, however, does not seem to be the case, since the terminal fields displayed a complex pattern of overlap and nonoverlap rather than a consistent shift of terminal fields expected in the case of preserved topology. The results of the present study are consistent with the view that pontine modules independently process information from different parts of individual cortical areas. We suggest that this characteristic property of the corticopontine projection system might be the morphological basis of the well established fact that somatotopically continuous sensory maps in the cortex are transformed into maps at the level of the cerebellar cortex, showing a fractured somatotopy.