Abstract: Objective To assess the validity of fasting plasma glucose (FPG) testing and hemoglobin A1c (HbA1c) detection in screening for diabetes (DM) in general population and to provide references for selection of DM screening method in epidemiolgoical survey.Methods FPG,HbA1c and postprandial 2-hour blood glucose (2hPG) detections were performed among 3 489 residents aged 18 years or above selected with stratified multistage cluster random sampling from 6 counties or districts covered by national vital registry system druing October to September 2010;optimal cut-point,sensitivity,specificity,and the area under curve of FPG and HbA1c related to DM were obtained with receiver operating characteristic (ROC) curve analysis.Results Based World Health Organization criteria 1999,2 975 of the residents (85.3%) had normal glucose tolerance (NGT);230 (6.6%) had impaired fasting glucose (IFG);176 (5.0%) had impaired glucose tolerance (IGT);and 108 (3.1%) were diagnosed as DM,respectively.The optimal cut-point value of FPG for the diagnosis of DM was 6.4 mmol/l;the value was associated with a sensitivity of 86.1% and a specificity of 96.5%;the area under the curve was 0.926 (95% confidence interval95%CI=0.887-0.965);the positive likelihood ratio (LR) was 24.60;the negative LR was 0.14;and the Youden index was 0.83.The optimal cut-point value of HbA1c for diagnosis of DM was 6.15%;the value was associated with a sensitivity of 66.7% and a specificity of 87.1%;the area under the curve was 0.814 (95%CI=0.762-0.867);the positive LR was 5.18 and the negative LR was 0.38;and the Youden index was 0.53.Conclusion FPG detection is a more reliable method for DM screening than HbA1c in epidemioligical survey.