1. Reinnervation of adult rat fourth deep lumbrical muscles was studied, following extensive partial denervation of the hindfoot by crushing the lateral plantar nerve (LPN). Most muscles remained innervated by between one and five motor axons supplied by the sural nerve (SN). Intact SN motor units expanded as a result of collateral sprouting. Virtually complete collateral reinnervation occurred in muscles containing more than two SN motor units. Twitch tension measurements from isolated muscles suggested that most of the sprouts evoked suprathreshold responses from the muscle fibres they innervated. Intracellular recordings suggested that only a small percentage of sprouts evoked subthreshold end-plate potentials. 2. Lateral plantar nerve motor axons returned to the lumbrical muscles within 15-18 days and subsequently reinnervated muscle fibres already innervated by SN motor nerve terminals. Nerve conduction in the regenerating axons was then blocked for 7-15 days by chronic superfusion with tetrodotoxin. In both LPN-blocked and control (LPN crushed but not blocked) animals, isometric tetanic tension overlap and intracellular recordings showed that some lumbrical muscle fibres became innervated exclusively by regenerating LPN motor axons. 3. With time, the tension evoked by stimulating regenerating motor axons increased and there was a parallel fall in the tension produced by stimulating the intact motor units. The extent of reinnervation by LPN motor axons was inversely related to the number of remaining SN motor units. In comparable muscles, regenerated LPN-blocked motor units produced only about half the tension of the controls. Selective glycogen depletion of motor units and intracellular recordings of end-plate potentials indicated that this was due to reduced numbers of muscle fibres innervated by the blocked motor axons. 4. Nerve conduction block prolonged the time course of the isometric twitch in regenerated motor units, and increased the duration of the end-plate potential in muscle fibres innervated only by the regenerating axons. LPN block did not affect the recovery of the latency of the end-plate potential. The regenerated motor units were more resistant to fatigue caused by continuous 4 Hz nerve stimulation than intact SN units, but the resistance to fatigue of LPN-blocked motor units was no different from the controls.(ABSTRACT TRUNCATED AT 400 WORDS)