The electrophysiological properties of 87 neurons in the deep dorsal horn (laminae III-VI) of the rat spinal cord have been investigated in vitro. Two preparations have been used; the transverse spinal cord slice preparation from the third or fourth lumbar segments of 14-16-day-old rats (71 cells) and a hemisected lumbar spinal cord preparation from 10-12-day-old rats (16 cells). The input impedances (range 11-128 M omega), membrane potentials (-67 +/- 8 mV S.D.), action potential amplitude (77 +/- 11.8 mV) duration (1.4 +/- 0.5 ms) and afterpotentials, were effectively identical in the neurons recorded from the two preparations. Neurons in both preparations when activated with long-duration (1-2 s) outward current pulses showed a single steady-state firing range with little adaptation of firing frequency or action potential amplitude. This pattern of responses was unaffected by changing the membrane potential. Orthodromic synaptic activity could be elicited in the neurons by stimulating either the small dorsal root remnants in the slice or the dorsal roots in the hemisected spinal cord. The responses evoked by single stimuli of increasing intensity varied in different neurons in both preparations. The commonest response (32/62) consisted of a short-latency, short-duration composite excitatory postsynaptic potential which generated one or two spikes with no further spiking activity at longer latency when the stimulus intensity was increased beyond threshold. In 20 neurons, graded stimulation produced a graded response with recruitment, at high intensities, of a discharge of action potentials lasting several hundred milliseconds. A small number of cells (4) responded to the single stimulus with a train of action potentials lasting several seconds. Stimulating adjacent dorsal roots in the hemisected cord preparation could evoke quite different responses from the neurons. The heterogeneity of the types of orthodromic responses obtained in both preparations, in spite of the almost uniform intrinsic membrane properties, is likely to reflect differences in the strength, location and type of afferent and interneuronal input to different dorsal horn cells.