1. The precentral bank of the Rolandic fissure of the cortical arm area has been explored with extracellular micro-electrodes in primates (baboons and monkeys) under nitrous oxide and oxygen anaesthesia, supplemented by small doses of Parkesernyl(R) and chloralose. The results in baboons and monkeys were the same.2. Single units were classified as pyramidal tract neurones or non-pyramidal tract neurones according to their antidromic responsiveness to stimuli applied in the dorsolateral funiculus at C1-2.3. Responses to electrical stimulation of the deep (motor) radial nerve, the deep palmar (motor) branch of the ulnar nerve, and the superficial (cutaneous) radial nerve could be recorded in the majority of neurones of the motor cortex provided that short trains of strong stimuli were used. Minimal responses to muscle nerve stimulation were observed in a few neurones at 1.4 x group I threshold, but most units reacted only with higher stimulus intensities (2-3 x group I threshold).4. The latencies to peripheral nerve stimulation were measured from the first peak of the incoming volley recorded at the root entry zone. The mean response latencies of pyramidal tract cells were between 20 and 25 msec; non-pyramidal tract cells were activated at slightly shorter mean latencies, the difference being significant for superficial radial nerve stimulation only (4 msec). These latencies are more than twice as long as those recorded in the postcentral gyrus, and the probability of discharge is lower than for postcentral neurones.5. A further difference between neurones of the postcentral and precentral gyrus is the pronounced convergence from different nerves and also from different modalities (cutaneous and muscle afferents) in units of the precentral cortex in contrast to units of the postcentral cortex.6. The high thresholds, necessary to activate precentral neurones by muscle nerve stimulation, make it unlikely that group I muscle afferents are involved. This is, furthermore, indicated by the lack of responsiveness to intravenous injection of succinylcholine which was, however, effective for driving neurones of the specific projection area for group I afferents, area 3a. The present experiments are consistent with the view that sensitivity of precentral neurones to muscle stretch (described in previous studies) is due to activation of secondary muscle spindle endings and their ascending pathways.7. The original hypothesis of a load compensating ;pyramidal reflex' with an oligosynaptic afferent contribution from the spindle primaries can be discarded. The present findings indicate that there is a feed-back from secondary muscle spindle afferents which, by way of a more complex pathway, can modulate the firing frequency of neurones in the motor cortex.