In some neurodegenerative diseases, genetic anticipation correlates with expansions of the CAG/CTG repeat sequence above the normal range through the generations of a pedigree. Among these neurodegenerative diseases are late onset autosomal dominant cerebellar ataxias (ADCA). ADCA are genetically heterogeneous disorders with different cloned genes for spinocerebellar ataxia type 1 (SCA1), type 2 (SCA2), type 3 or Machado-Joseph disease (SCA3/MJD), and type 6 (SCA6). Another related dominant ataxia, dentatorubral-pallidoluysian atrophy (DRPLA), also shows CAG/CTG repeat expansions. Genetic anticipation has been reported for all of them except for the recently cloned SCA6 gene. Other, as yet undetected SCA genes may show the same features. We have used the repeat expansion detection (RED) method to detect repeat expansions directly in DNA samples from ADCA patients not resulting from known genes. Our sample consists of 19 affected index cases, corresponding to 52.8% of our ADCA families without CAG/CTG repeat expansions in the SCA1, SCA2, SCA3/MJD, SCA6, or DRPLA genes. Eighty-nine percent of the index cases had expansions of a CAG/CTG sequence greater than 40 repeats by RED, while these were observed in only 26.9% of 78 healthy subjects from the general population (p < 0.0001). The distribution of RED fragments in controls and ADCA patients also shows significant differences with the Mann-Whitney U test (U = 376.5, p = 0.0007). Moreover, there was a significant inverse correlation between the size of expansion and the age of onset (r = -0.54, p = 0.018). These results show CAG/CTG repeat expansions of over 40 repeats in our sample of ADCA families not resulting from known SCA genes.