Individual rat embryos were chronically paralysed during development in utero by insertion of small slow-release capsules containing tetrodotoxin (TTX). In others, motoneurons were destroyed by individual injections of <latex>$\beta$</latex>-bungarotoxin (<latex>$\beta$</latex>-BTX). These embryos were compared with normal controls so that the effects of innervation and of use on embryonic development of skeletal muscles could be defined. The use of Sirius Red to stain muscle basal lamina made it possible to count muscle units (one or more myotubes contained within a single basal lamina sheath) with light microscopy. (i) In left hemidiaphragm muscles from control embryos about 20% of the adult number of muscle units was present on day 17 of gestation, the time at which formation of secondary myotubes was first observed, and about 70% at birth. New units continued to form during the first postnatal week. (ii) Skeletal muscles still formed following destruction of muscle innervation at day 14 of gestation, although at day 21 of gestation (birth) they were smaller and thinner than controls. The number of units in these muscles remained constant from day 18 through day 21 of gestation, and was similar to that in day 17 controls. No secondary myotubes were seen histologically. (iii) Following destruction of motoneurons at day 16, 17 1/2 or 18 of gestation, new muscle units continued to form for about 1 1/2 d, indicating that the temporary presence of innervation affected later development of muscle tissue. The denervated muscles contracted spontaneously during this time. (iv) TTX-induced paralysis from day 16 onwards was at least as effective as denervation in inhibiting the generation of new muscle units. There was a dose-response relation in the ability of TTX to reduce the generation of new muscle units beyond the amount necessary to paralyse nerves, and it is suggested that the higher concentrations also blocked myogenic contractions. High doses of TTX stopped generation of new units within half a day. (v) Even the slightest degree of recovery from paralysis was accompanied by restoration of the proper number of muscle units. This shows that there is no `critical period' for muscle fibre development within the last five prenatal days in the rat. (vi) It is concluded that development of primary myotubes is autonomous, but that development of secondary myotubes, which form the majority of muscle cells in the adult, is dependent on innervation. Even when nerves are present, new muscle fibres cannot be generated in the absence of muscle contraction.