It has been suggested that the mumps virus may be involved in the etiopathogenesis of Type-I diabetes mellitus. Most studies have analyzed this relationship retrospectively. We, however, carried out a prospective study over a 4 year period after a mumps infection in two age groups (16 years and under [group A no = 32] and over 16 years [group B no = 18]). These subjects with "diabetic risk factors" (impaired glucose tolerance, low insulin response, ICSA and/or HLA-DR3/DR4) were selected from 1581 registered cases, in whom an antecedent mumps infection had occurred in 1980 and 1981. Glucose tolerance and insulin secretion did not change significantly during 4 years after a mumps infection. Overt diabetes was not observed in any of the cases. One year after a mumps infection 35% of children and 63% of adolescents/adults exhibited ICSA (control subjects = 5%; a serum was considered ICSA-positive if more than 25% of the intact rat islet cells showed distinct cell surface immunofluorescence). After 4 years the percentage of subjects with ICSA decreased significantly to 13% and 14%, resp. Only 21% of ICSA-positive sera were found to be cytotoxic on rat islet cells (51Cr-release assay). No relationship could be evaluated between complications resulting from a mumps infection and the appearance of ICSA. There was no correlation between ICSA, glucose tolerance, and insulin secretion. In fact, our prospective study did not reveal any relationship between a mumps infection and Type-I diabetes. ICSA would seem to be of no predictive value.