Background Spinocerebellar ataxias are dominantly inherited neurodegenerative diseases. As potential treatments for these diseases are being developed, precise knowledge of their natural history is needed. We aimed to study the long-term disease progression of the most common spinocerebellar ataxias: SCA1, SCA2, SCA3, and SCA6. Furthermore, we aimed to establish the order and occurrence of non-ataxia symptoms, and identify predictors of disease progression. Methods In this longitudinal cohort study (EUROSCA), we enrolled men and women with positive genetic testing for SCA1, SCA2, SCA3, or SCA6 and with progressive, otherwise unexplained ataxia who were aged 18 years or older from 17 ataxia referral centres in ten European countries. Patients were seen every year for 3 years, and at irregular intervals thereafter. The primary outcome was the scale for the assessment and rating of ataxia (SARA), and the inventory of non-ataxia signs (INAS). We used linear mixed models to analyse progression. To account for dropouts, we applied a pattern-mixture model. This study is registered with ClinicalTrials.gov, number NCT02440763. Findings Between July 1, 2005, and Aug 31, 2006, 526 patients with SCA1, SCA2, SCA3, or SCA6 were enrolled. We analysed data for 462 patients with at least one follow-up visit. Median observation time was 49 months (IQR 35–72). SARA progression data were best fitted with a linear model in all genotypes. Annual SARA score increase was 2·11 (SE 0·12) in patients with SCA1, 1·49 (0·07) in patients with SCA2, 1·56 (0·08) in patients with SCA3, and 0·80 (0·09) in patients with SCA6. The increase of the number of non-ataxia signs reached a plateau in SCA1, SCA2, and SCA3. In patients with SCA6, the number of non-ataxia symptoms increased linearly, but more slowly than in patients with SCA1, SCA2, and SCA3 (p<0·0001). Factors that were associated with faster progression of the SARA score were short duration of follow-up (p=0·0179), older age at inclusion (0.04 [SE 0·02] per additional year; p=0·0476), and longer repeat expansions (0·06 [SE 0·02] per additional repeat unit; p=0·0128) in SCA1, short duration of follow-up (p<0·0001), lower age at onset (–0·02 [SE 0·01] per additional year; p=0·0014), and lower baseline SARA score (–0·02 [SE 0·01] per additional SARA point; p=0·0083) in SCA2, and lower baseline SARA score (–0·03 [SE 0·01] per additional SARA point; p=0·0195) in SCA6. In SCA3, we did not identify factors that affected progression of the SARA score. Interpretation Our study provides quantitative data on the progression of the most common spinocerebellar ataxias based on a follow-up period that exceeds those of previous studies. Our data could prove useful for sample size calculation and patient stratification in interventional trials. Funding EU FP6 (EUROSCA), German Ministry of Education and Research (BMBF; GeneMove), Polish Ministry of Science, EU FP7 (NEUROMICS).