The term spinal muscular atrophy (SMA) is used to encompass a group of inherited disorders in which the striking pathological feature is loss of the cell bodies of alpha motor neurons in the anterior horn cell of the spinal cord and, in some cases, of the bulbar motor nuclei. Although the pathological features of these disorders have been well characterized, the nature of the primary underlying biochemical abnormality remains to be determined. In the 1990s genetic linkage was established for the childhood onset recessive forms of SMA (types I, II and III) to markers mapping to the chromosomal region 5q11.2-13.3. Physical maps of the region were then constructed, several candidate genes isolated and in 1995 deletions in two genes, the survival motor neuron (SMN) gene and the neuronal apoptosis inhibitory protein (NAIP) gene, were identified in significant numbers of patients. Already the impact of the characterization of these deletions is being seen in clinical practice in terms of aiding diagnosis in symptomatic cases and in prenatal diagnosis. As discussed in this review however, several questions remain unresolved. It is unclear whether deletions in one or both of these genes, or indeed in other, as yet unidentified, genes are important in generating the SMA phenotype. The function of the protein product of the SMN gene is unknown. The NAIP gene encodes a protein which inhibits apoptosis in a mammalian cell line: is it disruption of this function which is relevant in SMA? What underlies the variation in disease severity evident both between and within families? Resolution of such issues is of crucial importance if the identification of these deleted gene sequences is to lead to the development of rational therapies for motor neuron diseases.