Seventy cases of primary degenerative cerebellar ataxias in ethnic Bengalees from southern West Bengal, India, were studied by the authors. Of these, 50 cases were of the familial type (hereditary ataxias) encountered in 23 families and the remaining 20 were of sporadic onset. 18 cases (from 11 families) were of "probable" autosomal recessive (AR) inheritance, 12 cases (8 families) had Friedreich's type ataxia (FA), 4 cases (2 families) had FA type ataxia with retained reflexes and in 2 cases (1 family) the exact phenotypic characterization could not be made. AR inheritance in these cases seemed most likely in view of the occurrence in a single generation with unaffected parents and history of consanguinity in many of the families studied. Genotypic confirmation of FA type ataxia and its variants could not be done in any case due to the non-availability of technology for studying the FA locus but some common dominant ataxia genotypes could be excluded. Thirty-two cases (from 12 families) with autosomal dominant ataxias (ADCA) were studied. Genotype analysis revealed 4 families with SCA2 genotype, 5 families with SCA3 and 3 families where genotypic characterization could not be made (phenotypically 2 were of ADCA I and 1 of ADCA II). No clear preponderance of one particular genotype of SCA over another could be demonstrated in our ethnic Bengalee patients. We also noted significant intra and inter-family variations in phenotypes within the same genotypic form as well as overlapping of clinical signs between different genotypes. Slow saccades and peripheral neuropathy were not seen consistently in our ethnic Bengalee subjects with SCA2 genotype. Similarly, extrapyramidal features, ophthalmoplegias and distal amyotrophy were seen in some but not all families with the SCA3 genotype. Phenotypic expression appeared to be an inconsistent marker of the SCA genotype in our patients. Of the 20 sporadic cases with cerebellar ataxia, genotype analysis revealed 2 cases with SCA1 and 1 with SCA2. Some of the sporadic ataxia cases had extracerebellar involvement and may warrant classification as Multiple System Atrophy. In all the 3 subjects with genotype characterization, phenotype correlation was lacking. The clinical pattern of hereditary ataxias in ethnic Bengalees seems to be somewhat different from that seen in Western India. The need for clinical and genetic studies of ataxias in different specific ethnic populations of India has been stressed.