IT has long been considered a general rule for nerve regeneration that the reinnervation of skeletal muscle is nonselective. Regenerating nerve fibers are supposed to reconnect with one skeletal muscle as readily as another according to studies covering a wide range of vertebrates (Weiss, 1937; Weiss & Taylor, 1944; Weiss & Hoag, 1946; Bernstein & Guth, 1961; Guth, 1961, 1962, 1963). Similarly, in embryogenesis proper functional connexions between nerve centers and particular muscles are supposedly attained, not by selective nerve outgrowth but rather through a process of 'myotypic modulation' (Weiss, 1955) that presupposes nonselective peripheral innervation. Doubt about the general validity of this rule and the concepts behind it has come from a series of studies on regeneration of the oculomotor nerve in teleosts, urodeles, and anurans and of spinal fin nerves in teleosts (Sperry, 1946, 1947, 1950, 1965; Sperry & Deupree, 1956; Arora & Sperry, 1957a, 1964). All point to the existence of basic selectivities in the formation of nerve-muscle connexions that vary in degree in different systems and in different species. Some 8 years ago we observed that complete section of the oculomotor nerve in cichlid fishes is soon followed by a full recovery or orderly and normally-patterned eye movements even when the nerve trunk is roughly teased apart with inevitable scrambling of the heterogeneous fibers of this nerve that supplies four of the extrinsic muscles of the eye (Arora & Sperry, 1957a). The following is an account of these and subsequent experiments that have been carried out in the interim in an effort to determine what kinds of factors are responsible for this extraordinary recovery that was taken initially to be another example of 'myotypic modulation'. The present findings point instead to an explanation in terms of selective peripheral innervation.