Efforts to classify the hereditary ataxias by their clinical and neuropathological phenotypes are troubled by excessive heterogeneity. Linkage analysis opened the door to a new approach with the methods of molecular biology. The classic form of autosomal recessive ataxia, Friedreich's ataxia (FA), is now known to be due to an intronic expansion of a guanine-adenine-adenine (GAA)-trinucleotide repeat. The autosomal dominant ataxias such as olivopontocerebellar atrophy (OPCA), familial cortical cerebellar atrophy (FCCA), and Machado-Joseph disease (MJD) have been renamed the spinocerebellar ataxias (SCA). Specific gene loci are indicated as SCA-1, SCA-2, SCA-3, SCA-4, SCA-5, SCA-6, and SCA-7. In 5 of them (SCA-1, SCA-2, SCA-3, SCA-6, and SCA-7), expanded cytosine-adenine-guanine (CAG)-trinucleotide repeats and their abnormal gene products cause the ataxic condition. The most common underlying loci for olivopontocerebellar atrophy (OPCA) are SCA-1 and SCA-2, although other genotypes may be added in the future. A major recent advance was the identification of the gene for SCA-3 and MJD, and the high prevalence of this form of autosomal dominant ataxia. In FA and the SCA with expanded CAG-trinucleotide repeats, clinical and neuropathological severity are inversely correlated with the lengths of the repeats. Anticipation in the dominant ataxias can now be explained by lengthening of the repeats in successive generations. Progress is being made in the understanding of the pathogenesis of FA and SCA as the absent or mutated gene products are studied by immunocytochemistry in human and transgenic murine brain tissue. In FA, frataxin is diminished or absent, and an excess of mitochondrial iron may cause the illness of the nervous system and the heart. In SCA-3, abnormal ataxin-3 is aggregated in neuronal nuclei, and in SCA-6, a mutated alpha1A-calcium channel protein is the likely cause of abnormal calcium channel function in Purkinje cells and in the death of these neurons.