1. Competition between two foreign nerves innervating frog skeletal muscle has been studied by using pairs of somatic motor nerves (s.m.n.s) or one s.m.n. and the preganglionic splanchnic nerve (s.p.n.) implanted into a denervated sartorius muscle that has been transplanted to the lymph sac of the back. 2. A single s.m.n. implanted into the muscle succeeded in innervating essentially every fibre within 2--3 months; tetanic stimulation of the nerve elicited 9--100% of the maximal direct tetanus tension. Most of the e.p.p.s were suprathreshold, since a single indirect stimulus evoked a twitch 60--100% as large as that to a direct stimulus. 3. If two s.m.n.s were implanted simultaneously, tetanic stimulation of either elicited 80--100% of the maximal tension to direct stimulation. If one nerve was implanted 2--3 months before the other, the second, although usually less effective than the first, normally innervated 50--100% of the fibres, with approximately the same time course of innervation as a single s.m.n. 4. Mutual synaptic repression was seen on examination of twitch tensions. With either simultaneous or staggered innervation, stimulation of each s.m.n. resulted in a twitch of 30--50% of the total direct twitch tension, with little overlap between the fields driven by the two nerves. Intracellular recordings showed that the distribution of subthreshold and spike-producing e.p.p.s reflected the existence of separate twitch fields. Even if one s.m.n. was implanted several months before the other and had time to establish suprathreshold junctions on most muscle fibres, an s.m.n. implanted later was able to reduce sharply the effectiveness of many junctions from the earlier nerve while itself innervating most muscle fibres. 5. The subthreshold e.p.p.s had low quantal content, typically ten or fewer quanta/e.p.p. The min e.p.p. frequency was very low, while min e.p.p. amplitude appeared to be normal. 6. In the vast majority of muscle fibres, junctions from the two nerves were not within recording distance of each other. Hence, we infer that the competitive interaction was mediated somehow via the muscle fibre. 7. The preganglionic splanchnic nerve, which also successfully reinnervated frog skeletal muscle, competed with a foreign s.m.n. in ways which differ qualitatively from the competition by a second s.m.n. In the presence of a s.m.n., synapses of the s.p.n. were almost universally subthreshold. However, if the s.p.n. was implanted 2--3 months before the s.m.n., the s.m.n. was prevented for several months from innervating fibres driven by the s.p.n. This delay in s.m.n. reinnervation was greater than if the first nerve implanted was also an s.m.n. 8. After 6--8 months of dual innervation by s.m.n. and s.p.n., the s.m.n. became almost totally dominant. However, if the s.m.n. was then sectioned, the s.p.n. became as effective, within approximately 1 week, as it would have been in the absence of the s.m.n.