Spinocerebellar ataxia type 31 (SCA31) is a recently defined subtype of autosomal dominant cerebellar ataxia (ADCA) characterized by adult-onset, pure cerebellar ataxia. The C/T substitution in the 5'-untranslated region of the puratrophin-1 gene (PLEKHG4) or a disease-specific haplotype within the 900-kb SCA31 critical region just upstream of PLEKHG4 has been used for the diagnosis of SCA31. Very recently, a disease-specific insertion containing penta-nucleotide (TGGAA)(n) repeats has been found in this critical region in SCA31 patients. SCA31 was highly prevalent in Nagano, Japan, where SCA31 accounts for approximately 42% of ADCA families. We screened the insertion in 94 SCA31 patients from 71 families in Nagano. All patients had a 2.6- to 3.7-kb insertion. The size of the insertion was inversely correlated with the age at onset but not associated with the progression rate after onset. (TAGAA)(n) repeats at the 5'-end of the insertion were variable in number, ranging from 0 (without TAGAA sequence) to 4. The number of (TAGAA)(n) repeats was inversely correlated to the total size of the insertion. The number of (TAGAA)(n) repeats was comparatively uniform within patients from the three endemic foci in Nagano. Only one patient, heterozygous for the C/T substitution in PLEKHG4, had the insertions in both alleles; they were approximately 3.0 and 4.3 kb in size. Sequencing and Southern hybridization using biotin-labeled (TGGAA)(5) probe strongly indicated that the 3.0-kb insertion, but not the 4.3-kb insertion, contained (TGGAA)(n) stretch. We also found that 3 of 405 control individuals (0.7%) had the insertions from 1.0 to 3.5 kb in length. They were negative for the C/T substitution in PLEKHG4, and neither of the insertions contained (TGGAA)(n) stretch at their 5'-end by sequencing. The insertions in normal controls were clearly detected by Southern hybridization using (TAAAA)(5) probe, while they were not labeled with (TGGAA)(5) or (TAGAA)(5) probe. These data indicate that control alleles very rarely have a nonpathogenic large insertion in the SCA31 critical region and that not only the presence of the insertion but also its size is not sufficient evidence for a disease-causing allele. We approve of the view that (TGGAA)(n) repeats in the insertion are indeed related to the pathogenesis of SCA31, but it remains undetermined whether a large insertion lacking (TGGAA)(n) is nonpathogenic.