Spinocerebellar ataxia type 1 (SCA1) is an autosomal dominant, late-onset neurodegenerative disease caused by a polyglutamine (polyQ) expansion in the ataxin-1 protein, which causes progressive neurodegeneration in cerebellar Purkinje cells and brainstem nuclei. Here, we tested if reducing mutant ataxin-1 expression would significantly improve phenotypes in a knock-in (KI) mouse model that recapitulates spatial and temporal aspects of SCA1. Adeno-associated viruses (AAVs), expressing inhibitory RNAs targeting ataxin-1, were injected into the deep cerebellar nuclei (DCN) of KI mice. This approach induced ataxin-1 suppression in the cerebellar cortex and in brainstem neurons. RNA interference (RNAi) of ataxin-1 preserved cerebellar lobule integrity and prevented disease-related transcriptional changes for over a year. Notably, RNAi therapy also preserved rotarod performance and neurohistology. These data suggest that delivery of AAVs encoding RNAi sequences against ataxin-1, to DCN alone, may be sufficient for SCA1 therapy.