Experiments were performed to determine whether thalamic neurons send collaterals to more than one subdivision of postcentral somatosensory cortex, and to evaluate the fidelity of somatotopic connections between cortical representations of the body surface. After microelectrode recordings identified the locations of the representations of the fingertips along the border of area 3b with area 3a and along the border of area 1 with area 2, one tracer, tritiated N-acetyl wheatgerm agglutinin ([3H]WGA), was injected into the rostral representation at the 3a-3b border, and another, wheatgerm agglutinin conjugated to horseradish peroxidase (WGA:HRP), was injected into the caudal representation at the 1-2 border. The results indicate that three thalamic regions are interconnected with both border zones. Regions of overlapping retrograde and anterograde label included the ventromedial portion of the lateral division of the ventroposterior nucleus (VPL), which is known to represent the digits; the medial extent of a dorsal capping zone of VPL that we term the ventroposterior superior nucleus (VPS); and the anterior pulvinar (Pa). Cells labeled with one or the other tracer, as well as cells labeled with both tracers, were found in VPL, VPS, and Pa. Our procedures necessarily underestimated the numbers of cells projecting to both rostral and caudal injection zones, but at least 15-20% of VPL and 20-40% of VPS neurons that were autoradiographically labeled appeared to have such collaterals. While only a few cells in Pa were autoradiographically labeled from injections at the 3a-3b border, the presence of several double-labeled cells suggests that at least some Pa neurons project to both injected zones. By relating the present results to those from studies where connections were determined for single architectonic fields, we conclude that many neurons in VP project to both areas 3b and 1, and that many neurons in VPS project to both areas 3a and 2. In addition, both areas 3a and 2 appear to have interconnections with Pa. Connections of Pa with area 2 have only recently been reported (Pons and Kaas, in press), and connections of Pa with area 3a have not been described. The results imply that, in part, the same information from cutaneous receptors is relayed to areas 3b and 1, and the same information from deep receptors is relayed to areas 3a and 2.(ABSTRACT TRUNCATED AT 400 WORDS)