Epidemiological data suggesting an infectious origin of diabetes pre-date the discovery of insulin; indeed it was the variation in mortality rates from diabetes that led Gunderson to hypothesise that a virus with 'selective affinity for the pancreas' may cause 'acute diabetes' in youth (1). He noted an increase in deaths from diabetes in young people aged 10-20 yr in Norway from 1900 to 1921 following epidemics of parotitis, with a lag time of 3-4 yr between infection and death. In Norway, Denmark,France, and America, the increase in deaths from diabetes exceeded the expected number based on population growth; lending further weight to the proposal that diabetes was caused by infection. Since that time,a large body of epidemiological, clinical and experimental research, in humans, cellular and animal models, has provided further insights into the contribution of infections in the development of type 1 diabetes.Epidemiological evidence for a viral aetiology of diabetes A substantial body of epidemiological data point to a significant contribution of the environment in the development of type 1 diabetes,although much of the evidence is not specific to viruses per se. These data include rising rates of type 1 diabetes in both developed and developing countries in recent decades (2, 3) and a reduced contribution of high risk human leucocyte antigen (HLA) genotypes (4, 5), indicating that non-genetic factors are important. Similarly, the pairwise concordance between monozygotic twins for type 1 diabetes of less than 40%, and the observation that the incidence of diabetes in migrant children reflects that of their adopted country (6, 7), provide circumstantial evidence that environmental agents contribute to the disease. Space-time clustering in the presentation of type 1 diabetes (8-10) and clustering of births in children who subsequently develop diabetes (11) support a direct role for infections in the initiation and acceleration of the disease process.