Spinal muscular atrophy (SMA) is a common human inherited disease characterized by degeneration of motoneurons and muscular atrophy. SMA results from deletions or mutations of the SMN (survival motor neuron) gene. A nerve-muscle coculture model, consisting of human muscle cells innervated by rat embryonic spinal cord explants, was used to study the pathogenesis of SMA. Previous studies have shown that myotubes formed by fusion of satellite muscle cells from patients with SMA I or SMA II (but not SMA III) underwent a characteristic degeneration 1–3 weeks after innervation. To correlate this cellular study with a molecular approach, we used reverse transcriptase–polymerase chain reaction (RT–PCR), and showed that SMN mRNAs were expressed throughout the fusion of normal satellite muscle cells with two peaks, the first appearing prior to the onset of fusion and the second one or two days before innervation. When satellite muscle cells from patients with SMA I or II were used, only the first peak was observed. Because in these cases the SMN telomeric gene (SMNtel) is deleted, it was concluded that the contribution of SMNtel-dependent mRNAs to the second peak is predominant in normal myogenesis and involved in maturation of myotubes. In addition, diseased satellite muscle cells did not fuse at the same rate as normal satellite muscle cells. Studies on myf-5, a muscle specific transcription factor family, showed that its expression was impaired during the fusion of satellite muscle cells from patients with SMA I or II compared with normal satellite muscle cells. Taken together, these observations suggest that (a) there is a muscle specific expression pattern of SMN, and (b) SMN probably plays a crucial role in maintenance of a functional motor unit, by allowing muscle cells to correctly differentiate and to allow motoneuron survival.