This study aimed to compare plasma insulin concentrations across the age-range from childhood to old age in the populations of Nauru and Tuvalu, and to assess their relationship to the incidence of impaired glucose tolerance and diabetes in young Nauruans. The studies, performed in 1975 and 1976, found that Nauru had a higher prevalence of Type 2 (non-insulin-dependent) diabetes mellitus than Tuvalu. Both studies included subjects of 8-29 years of age (n = 320 in Nauru, n = 318 in Tuvalu) and on these subjects glucose tolerance status, body mass index and fasting and 2-h (post 75 g glucose load) plasma insulin concentrations were determined. In Nauru, follow-up surveys in 1982 and 1987 included many of the subjects first seen in 1975/1976, allowing the incidence and natural history of glucose intolerance to be studied. Within the group of subjects with normal glucose tolerance, there was no effect of age on plasma insulin distributions in either population. However, in both populations, 8-19 year old subjects with normal glucose tolerance had higher body mass index-adjusted geometric mean fasting and 2-h insulin concentrations than older age-groups (p less than 0.001 for fasting insulin). Body mass index-adjusted geometric mean 2-h plasma insulin was higher in subjects with abnormal glucose tolerance relative to those with normal glucose tolerance in both populations. In Nauruans, 2-h insulin levels at baseline were predictive of impaired glucose tolerance and Type 2 diabetes in 1982, and fasting and 2-h insulin levels predicted development of Type 2 diabetes in 1987.(ABSTRACT TRUNCATED AT 250 WORDS)