In the search for the most efficacious antisense oligonucleotides (AOs) aimed at inducing SMN2 exon 7 inclusion, we systematically assessed three AOs, PMO25 (-10, -34), PMO18 (-10, -27), and PMO20 (-10, -29), complementary to the SMN2 intron 7 splicing silencer (ISS-N1). PMO25 was the most efficacious in augmenting exon 7 inclusion in vitro in spinal muscular atrophy (SMA) patient fibroblasts and in vitro splicing assays. PMO25 and PMO18 were compared further in a mouse model of severe SMA. After a single intracerebroventricular (ICV) injection in neonatal mice, PMO25 increased the life span of severe SMA mice up to 30-fold, with average survival greater by 3-fold compared with PMO18 at a dose of 20 μg/g and 2-fold at 40 μg/g. Exon 7 inclusion was increased in the CNS but not in peripheral tissues. Systemic delivery of PMO25 at birth achieved a similar outcome and produced increased exon 7 inclusion both in the CNS and peripherally. Systemic administration of a 10-μg/g concentration of PMO25 conjugated to an octaguanidine dendrimer (VMO25) increased the life span only 2-fold in neonatal type I SMA mice, although it prevented tail necrosis in mild SMA mice. Higher doses and ICV injection of VMO25 were associated with toxicity. We conclude that (1) the 25-mer AO is more efficient than the 18-mer and 20-mer in modifying SMN2 splicing in vitro; (2) it is more efficient in prolonging survival in SMA mice; and (3) naked Morpholino oligomers are more efficient and safer than the Vivo-Morpholino and have potential for future SMA clinical applications.