1. Extensor digitorum longus IV (EDL IV) of Xenopus laevis was isolated together with its nerve and the 8th and 9th spinal roots. Motor units were functionally isolated.2. Glass micro-electrodes were used to penetrate the muscle fibres and to record their electrical responses to stimulation of single motor axons, or the total motor supply to the muscle.3. Three types of electrical response were recorded from muscle fibres: an action potential in response to a single nerve stimulus (type 1), an action potential following repetitive nerve stimulation but not in response to single shocks (type 2) and end-plate potentials which summed to a plateau of depolarization during repetitive nerve stimulation without producing action potentials (type 3).4. In general muscle fibres giving type 1 responses belonged to fast motor units with high twitch: tetanus ratios; those giving type 3 responses, to small units with low twitch: tetanus ratios; and those giving type 2 responses, to units with intermediate properties.5. In many cases the several muscle fibres innervated by a single axon gave different responses to stimulation of that axon and a single muscle fibre might give different responses to stimulation of two single axons. Thus it was not possible to classify motor units or muscle fibres by electrical responses.6. Stimulation of the total motor supply to the muscle favoured the production of type 1 responses so that the proportion of muscle fibres giving type 1 responses was higher when the muscle nerve was stimulated than when single motor axons were stimulated. The proportion of fibres giving type 2 responses was lower, and the proportion of fibres giving type 3 responses was similar, in both cases. The implications of this are discussed.7. Low neuromuscular junctional efficacy is suggested as an explanation for fluctuating twitch tensions and low twitch: tetanus ratios in many motor units in this muscle, and may also affect the activity patterns imposed on the muscle fibres by the motor axons which supply them.