The study was undertaken to compare the effect of ADA and WHO criteria for screening of diabetes mellitus (DM) and intermediate glucose abnormality (Impaired fasting glucose/Impaired glucose tolerance-IFG/IGT) and to explore an acceptable fasting cut-off in a population-based study. Ten suburb villages with a population of 11,895 were selected purposively. Of the total 6235 eligible (> or = 20y) subjects, 4144 volunteered. We took height, weight, hip- and waist-girth, blood pressure and fasting blood glucose (FBG). All participants were classified into Group-1 (Gr-1: n=453) and Group-2 (Gr-2: n=3691), based on FBG above and below 5.4 mmol/l, respectively. All from Gr-1 and 610 randomized subjects from Gr-2 were investigated for oral glucose tolerance test (OGTT), HbA1c and lipids. The mean (SD) of age, body mass index (BMI) and FBG of all participants was 37.6 (15.2) y, 19.4 (2.9), and 4.7 (0.9) mmol/l, respectively. The prevalence of diabetes and IFG/IGT using American Diabetes Association (ADA) criteria were compared with WHO criteria separately in Gr-1 and Gr-2. For group-1, ADA criteria could diagnose 5.9% as diabetes and 2.1% as IFG, whereas, WHO criteria diagnosed 11.5% diabetes and 19% IGT. Likewise, in Gr-2, ADA detected much less than WHO criteria (DM: 0.3 vs. 2.3%; IFG/IGT 1.0 vs. 14.6%). We compared fasting and 2 hours post-load glucose (2-hBG) values according to percentiles. We found that 11.1 of 2-hBG corresponded with a fasting value that lies between 90 to 95th percentile, equivalent to 5.1-5.7 mmol/l. Using receiver operating characteristics (ROC) curve, we determined the cut-offs 4.6 - 5.4 mmol/l for IFG and > or = 5.5 for diabetes. Taking age and BMI into account the kappa agreements were better between the estimated cut-offs and the given 2-hBG values. The ADA cut-offs were found ineffective for screening. We proposed the modified fasting cut-offs for screening IFG and diabetes among the non-obese population.