1. In intact turtles, tactile stimulation of the body surface in the "shell pocket" region surrounding the hindlimb elicits a pocket scratch reflex, in which the hindlimb reaches toward and rhythmically rubs the stimulated site. In the present study, we utilized reduced in vitro preparations of the turtle spinal cord with attached peripheral nerves to investigate the time course and pharmacology of sensory-evoked excitability in the pocket scratch neural network. Fictive pocket scratch motor output was elicited by electrically stimulating either the ventral-posterior pocket (VPP) cutaneous nerve or the distal D8 (d.D8) nerve. Both nerves contain afferents innervating part of the pocket scratch receptive field. 2. Six-segment (D7-S2) preparations of the spinal cord, which included the entire hindlimb enlargement, produced fictive pocket scratch motor output in response to VPP nerve stimulation (n = 6). We recorded fictive motor output as electroneurograms from up to five peripheral nerves in D7-S2 preparations, including three knee extensor muscle nerves (IT-KE, which innervates triceps femoris pars iliotibialis; AM-KE, which innervates pars ambiens; and FT-KE, which innervates pars femorotibialis), a hip flexor (protractor) muscle nerve (VP-HP, which innervates puboischiofemoralis internus, pars anteroventralis), and a mixed cutaneous-muscle nerve that exhibits hip-extensor-correlated motor output during the scratch (d.D8). The timing characteristics of activity in these nerves during in vitro motor patterns were similar to what has been observed during the in vivo pocket scratch. 3. Even a single segment of spinal cord from the anterior hindlimb enlargement (D8) contained sufficient neural circuitry to generate rhythmic motor patterns in AM-KE, VP-HP, and d.D8 nerves during repeated stimulation of VPP (n = 5) or d.D8 (n = 1). Stimulus trains delivered at 3-5 Hz for > or = 6 s elicited one or more VP-HP bursts with clear burst terminations; in some cases, these were followed by distinct hip-extensor-correlated d.D8 bursts. AM-KE timing was characteristic of a pocket scratch synergy, beginning during the VP-HP burst and continuing after VP-HP offset. Thus even isolated D8 segments were capable of expressing rhythmic alternation between hip-flexor- and hip-extensor-correlated motor bursts as well as a pocket-scratch-specific knee-hip synergy. 4. A single electrical pulse delivered to the VPP or d.D8 nerve increased the excitability of the pocket scratch network in D7-S2 and D8 preparations for > or = 5-10 s. We estimated the time course of increased excitability by observing the temporal summation of scratch motor output in response to single pulses applied to cutaneous afferents at multisecond intervals. Stimulus parameters were adjusted so that a single pulse delivered to a "rested" preparation (rested = no stimulation for > 2 min) was at or just below threshold for evoking motor output. Single pulses delivered at 5- to 10-s intervals evoked strongly summating scratch motor output in D7-S2 and D8 preparations. These results show that neural mechanisms that store sensory-evoked excitation in the pocket scratch circuit exist within the spinal hindlimb enlargement and even within the isolated D8 segment. 5. With the use of in vitro preparations, we have begun to examine the pharmacology of sensorimotor processing in the pocket scratch network. Application of the N-methyl-D-aspartate (NMDA) receptor antagonist D-2-amino-5-phosphonovaleric acid (APV) (50-100 microM) to the spinal cord greatly reduced pocket scratch excitability. APV lowered the motor burst frequency of pocket scratch responses in D7-S2 preparations elicited by 3-Hz stimulation; it also reduced the amplitude of summating motor output in D7-S2 and D8 preparations in response to single electrical stimuli delivered at 5-s intervals. These results indicate that NMDA receptors have a key role in synaptic processing and sustained excitation within the pocket scratch neura