Résumé : OBJECTIVE: To identify factors that determine disease severity and clinical phenotype of the most common spinocerebellar ataxias (SCAs), we studied 526 patients with SCA1, SCA2, SCA3. or SCA6. METHODS: To measure the severity of ataxia we used the Scale for the Assessment and Rating of Ataxia (SARA). In addition, nonataxia symptoms were assessed with the Inventory of Non-Ataxia Symptoms (INAS). The INAS count denotes the number of nonataxia symptoms in each patient. RESULTS: An analysis of covariance with SARA score as dependent variable and repeat lengths of the expanded and normal allele, age at onset, and disease duration as independent variables led to multivariate models that explained 60.4% of the SARA score variance in SCA1, 45.4% in SCA2, 46.8% in SCA3, and 33.7% in SCA6. In SCA1, SCA2, and SCA3, SARA was mainly determined by repeat length of the expanded allele, age at onset, and disease duration. The only factors determining the SARA score in SCA6 were age at onset and disease duration. The INAS count was 5.0 +/- 2.3 in SCA1, 4.6 +/- 2.2 in SCA2, 5.2 +/- 2.5 in SCA3, and 2.0 +/- 1.7 in SCA6. In SCA1, SCA2, and SCA3, SARA score and disease duration were the strongest predictors of the INAS count. In SCA6, only age at onset and disease duration had an effect on the INAS count. CONCLUSIONS: Our study suggests that spinocerebellar ataxia (SCA) 1, SCA2, and SCA3 share a number of common biologic properties, whereas SCA6 is distinct in that its phenotype is more determined by age than by disease-related factors.