The fourth deep lumbrical muscle in the hind foot of adult rats was partially denervated by crushing the sural nerve (s.n.). The denervated muscle fibres became completely reinnervated by sprouts from lateral plantar nerve (l.p.n.) motor axons. By about 20 days after the nerve crush, s.n. motor axons started to reinnervate the muscle. In control muscles, a small proportion of the muscle fibres--about 2.5% of the muscle per motor unit--was reinnervated by s.n. motor axons over the following 20 days. Hence the regenerating terminals were able to re-establish functional synapses, despite the fact that all the muscle fibres were functionally innervated by l.p.n. terminals. When nerve impulse conduction in the l.p.n. was blocked with tetrodotoxin for up to 2 weeks, starting from the time when s.n. axons returned to the muscle, s.n. motor axons retrieved a much larger proportion of the muscle fibres--about 6.5% of the muscle per motor unit. There was a concomitant decrease in the tension produced by the sprouted l.p.n. motor axons. Intracellular recordings showed that many muscle fibres became innervated exclusively by regenerated s.n. motor nerve terminals. Measurements of end-plate potentials suggested that l.p.n. sprouts and the original nerve terminals were eliminated non-selectively. These results suggest that regenerating, active motor nerve terminals have an additional competitive advantage in reinnervating innervated muscles, if the intact terminals are inactive. When the l.p.n. was cut, rather than blocked, extensive reinnervation by the s.n. occurred-about 30% of the muscle per motor unit. This suggests that the absence of an intact nerve terminal in the motor end-plate provides a stronger stimulus than inactivity for synapse formation by regenerating motor axons.