1. Thirty-one physiologically identified primary afferent fibers were labeled intracellularly with horseradish peroxidase (HRP). 2. A computer analysis was used to determine whether the distribution of cutaneous mechanoreceptive afferent terminals varies as a function of location within the dorsal horn somatotopic map. 3. An analysis of the geometry of the projections of these afferents has shown that 1) terminal arbors have a greater mediolateral width within the region of the foot representation than lateral to it, 2) terminal arbors have larger length-to-width ratios outside the foot representation than within it, and 3) the orientation of terminal arbors near the boundary of the foot representation reflects the angle of the boundary. Previous attribution of mediolateral width variations to primary afferent type are probably in error, although there appear to be genuine variations of longitudinal extent as a function of primary afferent type. 4. Nonuniform terminal distributions represent the first of a three-component process underlying assembly of the monosynaptic portions of cell receptive fields (RFs) and the somatotopic map. The other two components consist of the elaboration of cell dendritic trees and the establishment of selective connections. 5. The variation of primary afferent terminal distributions with map location is not an absolute requirement for development of the map; for example, the RFs of postsynaptic cells could be assembled with the use of a uniform terminal distribution for all afferents, everywhere in the map, as long as cell dendrites penetrate the appropriate portions of the presynaptic neuropil and receive connections only from afferent axons contributing to their RFs.(ABSTRACT TRUNCATED AT 250 WORDS)