In Nembutal-anesthetized rats, 31 physiologically identified medullary dorsal horn (MDH) cells were labeled with horseradish peroxidase (HRP). Ten responded only to deflection of one or more vibrissae. Six cells were activated by guard hair movement only, six by deflection of guard hairs or vibrissa(e), and seven by pinch of facial skin with serrated forceps. Different classes of low-threshold cells could not be distinguished on the basis of their somadendritic morphologies or laminar distribution. Neurons activated by multiple vibrissae were unique, however, in that one sent its axon into the medial lemniscus, and three projected into the trigeminal spinal tract. None of the guard hair-only or vibrissae-plus-guard hair neurons had such projections. Cells that responded best to noxious stimulation were located mainly in laminae I, II, and deep V, while neurons activated by vibrissa(e) and/or guard hair deflection were located in layers III, IV, and superficial V. Low-threshold neurons generally had fairly thick dendrites with few spines, whereas high-threshold cells tended to have thinner dendrites with numerous spines. Moreover, the dendritic arbors of low-threshold cells were, for the most part, denser than those of the noxious cells. Neurons with mandibular receptive fields were located in the dorsomedial portion of the MDH; cells with ophthalmic fields were found in the ventrolateral MDH, and maxillary cells were interposed. Cells sensitive to deflection of dorsal mystacial vibrissae and/or guard hairs were located ventral to those activated by more ventral hairs. Neurons with rostral receptive fields were found in the rostral MDH, while cells activated by hairs of the caudal mystacial pad, periauricular, and periorbital regions were located in the caudal MDH. Receptive-field types were encountered that have not been reported for trigeminal primary afferent neurons: multiple vibrissae; vibrissae plus guard hairs; and wide dynamic range. The latter two can be explained by the convergence of different primary afferent types onto individual neurons. Our failure to find a significant relationship between dendritic area (in the transverse plane) and the number of vibrissae suggests that primary afferent convergence may not be responsible for the synthesis of the multiple vibrissae receptive field. Excitatory connections between MDH neurons may, therefore, account for multiple vibrissae receptive fields in the MDH.