The distribution, dendritic trees and axonal courses of spinally projecting cells in the dorsal column nuclei were studied after labelling by retrograde HRP transport. The region of densest distribution was at the base of the two nuclei and in the area between them, extending for about 2 mm caudally from the obex. Only very few cells were found inside the cell cluster regions of the nuclei, where their dendrites had a free stellate form. The great majority, lying between, deep, or rostral to the cluster regions, also had a stellate form, except where they impinged on the boundaries of the cluster regions or on other nuclear borders; the spread of dendrites was dramatically restricted at such boundaries, often leading to a fusiform appearance in transverse sections which however was not evident in the parasagittal plane. No justification was therefore found for subdividing the population on morphological grounds. Axons of these cells descended ipsilaterally in either the medial part of the dorsolateral fascicle or in the adjacent lateral part of the cuneate fascicle, at cervical levels, and probably in about equal numbers. Most axons destined for the DLF followed a deep caudolateral trajectory, while many destined for the DC had a more dorsal or lateral course. Collateral branches were seen within the nuclei but could not be followed far. The fact that few if any cells lying in the region of maximum distribution of the spinally projecting cells were labelled following injections of HRP into the thalamic ventroposterior nucleus emphasizes that they form a distinctive entity within this medullary nuclear complex, and that any axon branches they give into the contralateral brainstem must have some other destination than the VPL. Two other groups of neurons were labelled by HRP implants into the dorsal columns - one in the ventrolateral medullary reticular formation, and the other in the nucleus of the solitary tract.