Spinocerebellar ataxia type 31 (SCA31) is one of the most common forms of autosomal-dominant cerebellar ataxia in Japan. SCA31 has a strong founder effect, which is consistent with the fact that this disease is basically absent in other ethnicities. After searching the entire founder region of a 2-megabase (Mb), we finally identified a 2.5 to 3.8 kb-long complex penta-nucleotide repeat containing (TGGAA)n, (TAGAA)n, (TAAAA)n and (TAAAATAGAA)n as the only genetic change segregating SCA31 individuals from normal people. Furthermore, (TGGAA)n was isolated as the only repeat explaining the pathogenesis because other repeats were encountered in control Japanese. From the genomic point of view, the complex penta-nucleotide repeat lies in an intronic segment shared by two genes, BEAN1 (brain expressed, associated with Nedd4) and TK2 (thymidine kinase 2) transcribed in mutually opposite directions. While TK2 is ubiquitously expressed, BEAN1 is transcribed only in the brain. Thus, the complex repeat is bi-directionally transcribed exclusively in the brain, as two independent non-coding repeats. Furthermore, the complex repeat containing (UGGAA)n was found to form abnormal RNA structures, called RNA foci, in cerebellar Purkinje cell nuclei of SCA31 patients' brains. Subsequent investigation by over-expressing (UGGAA)n in Drosophila revealed that the RNA containing (UGGAA)n exerts toxicity in a length- and expression level-dependent manner, whereas its toxicity could be dampened by (UGGAA)n-binding proteins, TDP-43, FUS and hnRNP A2/B1. It seems rational to formulate a treatment strategy through enhancing the role of RNA-binding proteins against (UGGAA)n-toxicity in SCA31.