The dominant cerebellar ataxias (ADCAs) represent a clinically and genetically heterogeneous group of disorders linked by progressive deterioration in balance and coordination. The utility of genetic classification of the ADCAs has been highlighted by the striking variability in clinical phenotype observed within families and the overlap in clinical phenotype observed between those with different genotypes. The recent demonstration that spinocerebellar ataxia type 2 (SCA2) is caused by a CAG repeat expansion within the ataxin-2 gene has allowed us to determine the frequency of SCA2 compared with SCA1, SCA3/Machado-Joseph disease (MJD), and dentatorubropallidoluysian atrophy (DRPLA) in patients with sporadic and inherited ataxia. SCA2 accounts for 13% of patients with ADCA (without retinal degeneration), intermediate between SCA1 and SCA3/MJD, which account for 6% and 23%, respectively. Together, SCA1, SCA2, and SCA3/MJD constitute >40% of the mutations leading to ADCA I in our population. No patient without a family history of ataxia, or with a pure cerebellar or spastic syndrome, tested positive for SCA1, SCA2, or SCA3. No overlap in ataxin-2 allele size between normal and disease chromosomes, or intermediate-sized alleles, were observed. Repeat length correlated inversely with age at onset, accounting for approximately 80% of the variability in onset age. Haplotype analysis provided no evidence for a single founder chromosome, and diverse ethnic origins were observed among SCA2 kindreds. In addition, a wide spectrum of clinical phenotypes was observed among SCA2 patients, including typical mild dominant ataxia, the MJD phenotype with facial fasciculations and lid retraction, and early-onset ataxia with a rapid course, chorea, and dementia.