Spinocerebellar ataxia type 31 (SCA31) is an adult-onset autosomal-dominant neurodegenerative disorder showing progressive cerebellar ataxia mainly affecting Purkinje cells. The SCA31 critical region was tracked down to a 900 kb interval in chromosome 16q22.1, where the disease shows a strong founder effect. By performing comprehensive Southern blot analysis and BAC- and fosmid-based sequencing, we isolated two genetic changes segregating with SCA31. One was a single-nucleotide change in an intron of the thymidine kinase 2 gene (TK2). However, this did not appear to affect splicing or expression patterns. The other was an insertion, from 2.5-3.8 kb long, consisting of complex penta-nucleotide repeats including a long (TGGAA)n stretch. In controls, shorter (1.5-2.0 kb) insertions lacking (TGGAA)n were found only rarely. The SCA31 repeat insertion's length inversely correlated with patient age of onset, and an expansion was documented in a single family showing anticipation. The repeat insertion was located in introns of TK2 and BEAN (brain expressed, associated with Nedd4) expressed in the brain and formed RNA foci in the nuclei of patients' Purkinje cells. An electrophoretic mobility-shift assay showed that essential splicing factors, serine/arginine-rich splicing factors SFRS1 and SFRS9, bind to (UGGAA)n in vitro. Because (TGGAA)n is a characteristic sequence of paracentromeric heterochromatin, we speculate that the insertion might have originated from heterochromatin. SCA31 is important because it exemplifies human diseases associated with "inserted" microsatellite repeats that can expand through transmission. Our finding suggests that the ectopic microsatellite repeat, when transcribed, might cause a disease involving the essential splicing factors.