The cerebellar ataxia of Marie (SCA) was investigated in Western Norway, a region with a population of 725,000, and with several isolated communities. Two modes of transmission were found: autasomal dominant SCA segregating in four families, and autosomal recessive SCA segregating in nine families. Within the second category, two distinct clinical types were observed. One exhibited ataxia with additional spinal involvement, including both upper and lower motor neuron affection and sensory disturbances. In some cases there were also ophthalmoplegias, dementia, epilepsy, and Parkinsonian signs. The other type exhibited mostly ataxic signs with little additional spinal involvement, but with dementia, epilepsy, and ophthalmoplegias in some cases, and occasionally hyperkinesia and optic nerve atrophy. The autosomal dominant SCA cases showed a varied spino-cerebellar symptomatology, but lacked ocular signs and had less mental disturbances than recessive SCA. Additionally, the disease course was more benign in autosomal dominant than in autosomal recessive SCA. The dominant form did not affect life expectancy, whereas autosomal recessive SCA considerably reduced the life span.In four families displaying autosomal dominant SCA, 32 persons were examined: 11 were affected, 10 had unspecific neuropathy (Un), and 11 were unaffected. In six families with autosomal recessive SCA of the spino-cerebellar type, 53 persons were examined, of whom 11 were affected, 14 had Un, and 28 were unaffected. In two families (the third has been described elsewhere) with autosomal recessive SCA of the cerebellar type, 57 persons were examined: 12 were affected, 19 had Un, and 26 were unaffected. In the diagnosis of Un, a score system was used to record all types of neurological signs. The scores were corrected for effects of age and sex, based on findings in a normal population.The estimated prevalence figures in the region investigated were: 3.2/100,000 for autosomal dominant, 1.8/100,000 for spino-cerebellar autosomal recessive, and 1.2/100,000 for cerebellar autosomal recessive SCA; the gene frequencies were 3.9.10-5, 2.5.10-3, and 4.8.10-4, respectively. The distribution of Un in SCA families indicated different aetiologies. Ratios in the recessive SCA families were compatible with Un scmetimes representing a heterozygous SCA manifestation. The prevalence in dominant SCA families fitted with a hypothesis that Un behaved here as a polygenic trait. Clinical Un differences also supported these contentions. Un polygenes are assumed to be important contributors to the inter-and intrafamilial variation of phenotypes in autosomal dominant SCA.