Small interfering RNA (siRNA) holds therapeutic promise for silencing dominantly acting disease genes, particularly if mutant alleles can be targeted selectively. In mammalian cell models we demonstrate that allele-specific silencing of disease genes with siRNA can be achieved by targeting either a linked single-nucleotide polymorphism (SNP) or the disease mutation directly. For a polyglutamine neurodegenerative disorder in which we first determined that selective targeting of the disease-causing CAG repeat is not possible, we took advantage of an associated SNP to generate siRNA that exclusively silenced the mutant Machado–Joseph disease/spinocerebellar ataxia type 3 allele while sparing expression of the WT allele. Allele-specific suppression was accomplished with all three approaches currently used to deliver siRNA: in vitro-synthesized duplexes as well as plasmid and viral expression of short hairpin RNA. We further optimized siRNA to specifically target a missense Tau mutation, V337M, that causes frontotemporal dementia. These studies establish that siRNA can be engineered to silence disease genes differing by a single nucleotide and highlight a key role for SNPs in extending the utility of siRNA in dominantly inherited disorders.