Horseradish peroxidase was injected intra-axonally into functionally identified cutaneous primary afferent fibers in the cat's spinal trigeminal nucleus in order to study the morphology of their central terminations. They were determined physiologically to be large, myelinated primary afferents including vibrissa, G1-hair, and slowly adapting type I afferents, as well as small, myelinated primary afferents including D-hair and A-delta high-threshold mechanoreceptive (HTM) afferents. The axons were stained for distances of 4-12 mm at the levels of the subnuclei interpolaris (Vi) and caudalis (Vc) of the spinal trigeminal nucleus. The collaterals of large, myelinated primary afferents formed terminal arbors in the outer part of the spinal trigeminal nucleus when they were rostral to or near the obex (rostral-type collaterals), in lamina V when in the rostral part of Vc (caudalis-type collaterals), and in lamina III/IV when in the caudal part of Vc (spinal dorsal horn-type collaterals). There were no transitional forms between the rostral and caudalis types, but there was a transitional type which terminated in laminae III/IV and V between the caudalis and spinal dorsal horn types. Major morphological differences were not observed among the three functional types of large, myelinated primary afferents. The collaterals of D-hair afferents showed the same types of collaterals as were seen in the large, myelinated primary afferents, except that terminal arbors of the spinal dorsal horn-type collaterals were distributed in lamina IIb in addition to lamina III/IV. Collaterals of A-delta HTM afferents formed extensive terminal arbors in the superficial part of Vi. In Vc they were distributed mainly in lamina I and few terminals existed in lamina IIa. The frequency of collaterals of A-delta HTM afferents was less than that of the other types of afferents. The collaterals of A-delta HTM afferents in Vi were distributed less frequently than those in Vc. The terminal arbors of A-delta HTM afferents were less extensive than those of the other types of afferents. The average size of varicosities of A-delta HTM afferents was smaller and that of vibrissa afferents was larger than that of the other types of afferents. This study demonstrated that there are large differences in the terminations of nociceptive and non-nociceptive primary afferents and that Vi receives direct nociceptive input from facial skin. However, the sparse distribution of collaterals of HTM afferents in Vi suggests that the rostral trigeminal nucleus contributes partially to sensory processing of facial pain.