Skeletal muscles formed in the absence of innervation are deficient in number of fibers, atrophic, and prone to degeneration. We and others have suggested that primary myotubes form autonomously, and the major effect of embryonic denervation is to halt the formation of secondary myotubes by removing the mitotic stimulus for generation of their precursor myoblasts. This hypothesis has recently been put into question by the results of work by Condon et al. (Dev. Biol., 138, 275-295, 1990) and by B. J. Fredette and L. T. Landmesser (Dev. Biol. 143, 19-35, 1991) who propose that lowered secondary myotube numbers are at least in part due to absence of the primary myotubes which would normally serve as scaffold for their formation. We distinguish between these hypotheses by studying generation and degeneration of primary myotubes and formation of secondary myotubes in fetal rat muscles denervated by injection of beta-bungarotoxin at different times between Embryonic Days (E) 14 and 17. Denervation at any of these times reduced primary myotube numbers on E18 to about 80% of control. Despite this modest decrease in primary myotube numbers, secondary myotube generation fell to below 3% of normal in extensor digitorum longus (EDL) muscles denervated on E14 or E15 and to below 12% of control in soleus muscles denervated on E14-E16. Secondary myotubes were associated with degenerating as well as intact primary myotubes. More secondary myotubes were generated only if nerves had been present up until E16 for EDL or E17 for soleus. We suggest that these observations of a critical period until E16 (EDL) or E17 (soleus) during which innervation is an absolute prerequisite for secondary myotube formation reflect the presence of a nerve-dependent population of myoblasts which must be activated before secondary myotube formation can commence.