The molecular bases of autosomal dominant cerebellar ataxia (ADCA) have been increasingly elucidated, but 17-50% of ADCA families still remain genetically undefined in Japan. In this study we investigated 67 genetically undefined ADCA families from the Nagano prefecture, and found that 63 patients from 51 families possessed the -16C>T change in the puratrophin-1 gene, which was recently found to be pathogenic for 16q22-linked ADCA. Most patients shared a common haplotype around the puratrophin-1 gene. All patients with the -16C>T change had pure cerebellar ataxia with middle-aged or later onset. Only one patient in a large, -16C>T positive family did not have this change, but still shared a narrowed haplotype with, and was clinically indistinguishable from, the other affected family members. In Nagano, 16q22-linked ADCA appears to be much more prevalent than either SCA6 or dentatorubral-pallidoluysian atrophy (DRPLA), and may explain the high frequency of spinocerebellar ataxia.