Background Despite recent progress in the field of genetics, sporadic late-onset (> 40 years) cerebellar ataxia (SLOCA) etiology remains frequently elusive, while the optimal diagnostic workup still needs to be determined. We aimed to comprehensively describe the causes of SLOCA and to discuss the relevance of the investigations. Methods We included 205 consecutive patients with SLOCA seen in our referral center. Patients were prospectively investigated using exhaustive clinical assessment, biochemical, genetic, electrophysiological, and imaging explorations. Results We established a diagnosis in 135 (66%) patients and reported 26 different causes for SLOCA, the most frequent being multiple system atrophy cerebellar type (MSA-C) (41%). Fifty-one patients (25%) had various causes of SLOCA including immune-mediated diseases such as multiple sclerosis or anti-GAD antibody-mediated ataxia; and other causes, such as alcoholic cerebellar degeneration, superficial siderosis, or Creutzfeldt–Jakob disease. We also identified 11 genetic causes in 20 patients, including SPG7 (n = 4), RFC1 -associated CANVAS (n = 3), SLC20A2 ( n  = 3), very-late-onset Friedreich’s ataxia ( n  = 2), FXTAS ( n  = 2), SCA3 ( n  = 1), SCA17 ( n  = 1), DRPLA ( n  = 1), MYORG ( n  = 1), MELAS ( n  = 1), and a mitochondriopathy ( n  = 1) that were less severe than MSA-C ( p  < 0.001). Remaining patients (34%) had idiopathic late-onset cerebellar ataxia which was less severe than MSA-C ( p  < 0.01). Conclusion Our prospective study provides an exhaustive picture of the etiology of SLOCA and clues regarding yield of investigations and diagnostic workup. Based on our observations, we established a diagnostic algorithm for SLOCA.