Spinocerebellar ataxias comprise a poorly understood group of inherited degenerative neurological diseases. Attempts to classify hereditary ataxias on the basis of the neurological features or specific clinical signs such as tendon reflex changes have proven to be unsatisfactory. Early onset cerebellar ataxia (EOCA) is generally inherited as an autosomal-recessive trait. Thus far, we do not have accurate answers to several questions about its classification. However, significant clinical heterogeneity observed in four Tunisian families with typical EOCA clinical features reinforces the hypothesis of genetic heterogeneity underlying this phenotype. We have demonstrated that three of the four families studied were not linked to Friedreich's ataxia (FA), vitamin E deficiency ataxia (AVED), and autosomal dominant cerebellar ataxia (ADCA) loci. The fourth family showed homozygosity for a large pathological expansion of GAA repeat in all patients, the parents being heterozygous for this mutation. We have also noted, in the case of the family studied, that there was instability in the transmission of the mutation, along with a phenomenon of anticipation comparable to that observed in dominant triplet repeat diseases. EOCA is thus clinically indistinguishable from FA, yet genetically independent of all known candidate genes. Genetic mapping is required for research into the causal gene and an understanding of the disease's physiopathologic mechanisms.