Aim: Recently, a change of the diagnostic tool for diabetes from an oral glucose tolerance test (OGTT) to hemoglobin A1c (HbA1c) has been suggested. The aim of the study was to assess whether ethnicity modified the association between glucose levels and HbA1c and to compare diabetes prevalence according to diagnostic method among Greenland Inuit, Inuit migrants in Denmark, and a general Danish population. Methods: Data from two population-based surveys conducted from 1999-2002 were compared with a total of 7957 individuals, 1173 Inuit participants from the Greenland Population Study, including 256 Inuit migrants in Denmark, and 6784 Danish participants in the Inter99 study. The participants received a standard 75-g OGTT. HbA1c was analyzed by an ion-exchange HPLC Bio-Rad variant (Hercules, California). All analyses were performed in the laboratory at Steno Diabetes Centre. Results: The Inuit had significantly higher levels of HbA1c than the Danish participants at any given level of fasting and 2-h glucose and for each category of glucose tolerance: normal glucose tolerance, impaired fasting glycemia, and impaired glucose tolerance. The prevalence of diabetes diagnosed by OGTT was 11.2% among Inuit residents in Greenland, 9.8% among Inuit migrants, and 4.2% among Danes vs. 31.7% among Inuit residents in Greenland, 21.3% among Inuit migrants, and 6.7% among Danes diagnosed by HbA1c. OGTT-defined diabetes was associated with increased cardiovascular risk factors compared to HbA1c-defined diabetes. Conclusion: The association between glucose and HbA1c is not the same in the two populations. The mechanism behind this difference is unknown, and studies on long-term consequences associated with HbA1c are needed.