To evaluate the adequacy and usefulness of the stable glycated hemoglobin (HbA(1c)) value of 6.5% suggested by the Japan Diabetic Society in 1999 for supportive diagnostic marker of diabetes, we assessed the sensitivity and specificity of an HbA(1c) value of 6.5% in patients who were newly diagnosed by the 75 g oral glucose tolerance test (75g-OGTT). A total of 866 Japanese subjects underwent the 75g-OGTT and HbA(1c) measurement (normal range: 4.3-5.8%). They were divided into three groups [normal glucose tolerance (NGT), impaired glucose tolerance (IGT), and diabetes mellitus (DM)], using the WHO criteria, since no subject with impaired fasting glycemia (IFG) was observed. The cut-off value of HbA(1c) separating DM from NGT or DM from IGT on cumulative distribution curve analysis was 5.9% (sensitivity 0.76 and specificity 0.86) and 5.9% (sensitivity 0.76 and specificity 0.77), respectively. The sensitivity of an HbA(1c) of 6.5% for separation of DM from NGT or IGT by the same analysis was 0.49 and 0.49, respectively. Similarly, the specificity for separation of DM from NGT or IGT was 0.98 and 0.98, respectively. These results mean that 49% of diabetic subjects show an HbA(1c)> or =6.5%, and 51% have an HbA(1c) less than 6.5%, while only 2% of NGT and IGT subjects have an HbA(1c)> or =6.5%, and 98% have a value less than 6.5%. Therefore, the sensitivity of an HbA(1c) value of 6.5% in separating DM from NGT or IGT is low, and thus 6.5% is too high value to use when screening for diabetes. However, the specificity is very high, so an HbA(1c) of 6.5% is a useful supportive marker to diagnose diabetes.