BACKGROUND: Geographical variation in the distribution of variant Creutzfeldt-Jakob disease (vCJD) might indicate the transmission route of the infectious agent to man. We investigated whether regional incidences of vCJD were correlated with regional dietary data. METHODS: The National CJD Surveillance Unit prospectively identified 84 people with vCJD up to Nov 10, 2000, in Great Britain. Their lifetime residential histories were obtained by interviews with a close relative. Cumulative incidences of vCJD by standard region were calculated. Grid references for places of residence in 1991 were identified and evidence of geographical clusters were sought. Data on diet in the 1980s were analysed for regional correlations with vCJD incidence. The socioeconomic status of the places of residence of people with vCJD was compared with that of the general population. FINDINGS: vCJD incidence was higher in the north of Great Britain than the south. The rate ratio (north vs south) was 1.94 (95% CI 1.27-2.98). The mean Carstairs' deprivation score for areas of residence of people with vCJD was -0.09 (-0.73 to 0.55), which is close to the national average of zero. Regional rates of vCJD were correlated with consumption of other meat or meat products as classified and recorded by the Household Food Consumption and Expenditure Survey (r=0.72), but not with data from the Dietary and Nutritional Survey of British Adults. Five people with vCJD in Leicestershire formed a cluster (p=0.004). INTERPRETATION: Regional differences in vCJD incidence are unlikely to be due to ascertainment bias. We had difficulty determining whether regional variations in diet might cause these differences, since the results of dietary analyses were inconsistent.