Spinocerebellar ataxias (SCAs) are caused by expansion of (CAG)n triplet repeats. These repeats occur as polymorphic forms in general population; however, beyond a threshold size they become pathogenic. The sizes and distributions of repeats at the SCA1, SCA2, SCA3, SCA7 and DRPLA loci were assessed by molecular analysis of 124 unrelated ataxia patients and 44 controls, and the association of larger normal (LN) alleles with disease prevalence was evaluated. Triplet repeat expansions in the disease range were detected in 8% (10/124) of the cases, with the majority having expansion at the SCA1 locus. Normal allele ranges in the cohort studied were similar to the Caucasian and North Indian populations but differed from the Korean and Japanese populations at various loci. The percentage of individuals with LN alleles at the SCA1 and SCA2 loci was higher than reported in Indians, Japanese and Caucasians. LN alleles showed a good correlation with the incidence of SCA1, indicating that SCA1 is the most prevalent ataxia in our population. The majority of cases with clinical symptoms of SCA could not be diagnosed by established CAG repeat criteria, suggesting that there may be an alternative basis for disease pathogenesis: (i) Repeats lower than the normal range may also result in abnormal phenotypes (ii) LN alleles at different loci in the same individual may contribute to symptoms (iii) Exogenous factors may play a role in triggering disease symptoms in individuals with LN alleles (iv) Triplet repeats may reach the disease range in the brain but not in the blood.