Spinal muscular atrophy (SMA) is an autosomal recessive disease caused by mutation or deletion of the survival motor neuron gene 1 (SMN1). The highly homologous gene, SMN2, is present in all patients, but it cannot compensate for loss of SMN1. SMN2 differs from SMN1 by a few nucleotide changes, but a C --> T transition in exon 7 leads to exon skipping. As a result, most transcripts from the SMN2 gene lack exon 7. Although SMN1 is the disease-determining gene, the number of SMN2 copies appears to modulate SMA clinical phenotypes. Thus, determining the SMN copy number is important for clinical diagnosis and prognosis. We have developed a quantitative real-time TaqMan polymerase chain reaction assay for both the SMN1 and SMN2 genes, in which reliable copy number determination was possible on deoxyribonucleic acid samples obtained by two different isolation methods and from two different sources (human blood and skin fibroblasts). For SMN1, allele specificity was attained solely by addition of an allele-specific forward primer and, for SMN2, by addition of a specific forward primer and a nonextending oligonucleotide (SMN1 blocker) that reduced nonspecific amplification from SMN1 to a negligible level. We validated the reliability of this real-time polymerase chain reaction approach and found that the coefficient of variation for all the gene copy number measurements was below 10%. Quantitative analysis of the SMN copy number in SMA fibroblasts by this approach showed deletion of SMN1 and an inverse correlation between the SMN2 copy number and severity of the disease.