Spinal muscular atrophy (SMA) is an autosomal recessive neurodegenerative disorder caused by SMN1 gene mutations. About 40% of SMN1 subtle mutations produced premature termination codons (PTC). This study aims to determine the capacity of these PTCs to trigger nonsense-mediated mRNA decay (NMD) pathway.Three nonsense mutations in SMN1, including c.43C > T, c.683T > A and c.844C > T, were investigated by using a minigene system and in vivo splicing assays. Two strategies were supplied: administration of cycloheximide (NMD inhibitor) and knockdown of UPF1 (a key NMD factor) in the cells carrying different minigenes.The wild-type minigene exclusively produced correctly spliced transcripts (FL-SMN1). Both the 683T > A and 844C > T expressed remarkably lower FL-SMN1 than the wild-type cells. After cycloheximide treatment, the FL-SMN1 levels in both the 683T > A and 844C > T were increased significantly compared with that of untreated cells. UPF1 knockdown in both the mutant 683T > A and 844C > T caused a dramatically augmentation of FL-SMN1 as compared to that in the cells treated with non-specific control siRNAs.Our data provide evidence that c.683T > A and c.844C > T, but not c.43C > T, in SMN1 leading to SMA trigger NMD using a minigene system. Therefore, NMD should be taken into consideration when exploring the pathogenetic mechanisms for these mutations.