OBJECTIVES: To investigate the cause of sporadic motor neuron disease (MND) by twin study, so allowing (1) estimation of the genetic contribution, and (2) collection of matched pairs for a case-control study of possible environmental factors. METHODS: 10872 death certificates bearing the diagnosis MND were collected from 1979 to 1989 inclusive. Inspection of individual birth entries allowed identification of potential twins. The status of each co-twin was determined and contact made through the National Health Service Central Register (NHS-CR) and their general practitioner (GP). The diagnosis of MND was verified via the co-twin and relatives, and medical records where available. Zygosity was assessed using a recognised questionnaire. Details concerning environmental exposures and health were gathered by interview of cotwin and relatives using a semistructured questionnaire. Heritability (h2) of MND was estimated, and the environmental information was analysed by conditional logistic regression modelling. RESULTS: Seventy seven probands were identified, of whom 26 were monozygotic and 51 dizygotic. Four monozygotic probands were concordant, but two probands came from a family known to have familial MND. The estimated heritability was between 0.38 and 0.85. Most environmental risk factors were not significant. Regular vehicle maintenance (odds ratio (OR) = 7.0; 95% confidence interval (95% CI) 1.3-89.9) and occupational paint usage (OR = 3.75; 95% CI 1.0-17.1), however, occurred significantly more often in the affected cases. CONCLUSIONS: This "death discordant" method for twin collection has proved to be viable, and has allowed the ascertainment of a large population sample in a rare disease. The genetic role in sporadic MND is substantial, and higher than expected. Exposure to industrial chemicals, particularly constituents of petrochemicals and paints, may contribute to the aetiology of MND.