Administration of catechol to rats anaesthetized with urethane produces a central excitatory state during which an EMG consisting of three temporally distinct components (M1, M2 and M3) can be recorded from forelimb and hindlimb muscles to electrical stimulation of cutaneous afferents. The probability of occurrence of all three components was measured in flexor and extensor muscles of fore- and hindlimb and showed that the long latency component (M3) occurred less frequently in hindlimb muscles than forelimb and that its probability of occurrence in hindlimb extensors was significantly reduced as compared to flexors. A possible reflex pathway for this long latency component of the EMG is suggested. Phenobarbitone (5 mg kg-1 i.v.) had no significant effect on the probability of occurrence of M1, M2 or M3. These results suggest that the long latency component (M3) is not due to activation of a spino-bulbo-spinal reflex as has been previously suggested. Cholinoceptor blocking drugs were found to reduce significantly the probability of occurrence of M2 and M3 and anticholinesterases to increase the probability of M2. None of the drugs was found to affect the short latency M1 component of the EMG. These results are discussed in relation to the possible reflex pathways of all three components of the EMG.