Epidemiological studies in Pacific populations have suggested a relationship between glucose tolerance and proportional Austronesian genetic admixture, with non-Austronesian Melanesians relatively free of glucose intolerance. However, a survey conducted in 1985 demonstrated the apparent emergence of glucose intolerance in a peri-urban non-Austronesian community, casting doubt on this hypothesis. In 1986 glucose tolerance was studied in three village communities on Karkar Island, Papua New Guinea. Two were of Austronesian and one was of non-Austronesian genetic ancestry. Prevalence of diabetes was low in all communities. However, prevalence of impaired glucose tolerance (IGT) was 7% in the more developed Austronesian community, as compared with 2% in the equivalent non-Austronesian group (p less than 0.05). Multiple regression analysis demonstrated not only a highly significant association between 2-h plasma glucose and insulin concentrations (p less than 0.001), but also that for a given value of plasma insulin, 2-h plasma glucose values were lower for non-Austronesian than for Austronesian subjects (males, p less than 0.05, females, p less than 0.01). However, average plasma insulin concentrations were relatively high in the non-Austronesian community, despite lower average plasma glucose concentrations.