The efficacy of using fasting plasma glucose (FPG) alone as a preferred screening test for diabetes has been questioned. This study was aimed to evaluate whether the use of serum advanced glycation end products-peptides (sAGEP) would help to improve the efficacy of FPG in diabetes screening among high-risk Chinese subjects with FPG <7.0 mmol/L. FPG, 2-h plasma glucose (2h-PG), serum glycated haemoglobin A1c (HbA1c), and sAGEP were measured in 857 Chinese subjects with risk factors for diabetes. The areas under receiver operating characteristic (ROC) curves generated by logistic regression models were assessed and compared to find the best model for diabetes screening in subjects with FPG <7.0 mmol/L. The optimal critical line was determined by maximizing the sum of sensitivity and specificity. Among the enrolled subjects, 730 of them had FPG <7.0 mmol/L, and only 41.7% new diabetes cases were identified using the 1999 World Health Organization FPG criterion (FPG ≥7.0 mmol/L). The area under ROC curves generated by the model on FPG-sAGEP was the largest compared with that on FPG-HbA1c, sAGEP, HbA1c or FPG in subjects with FPG <7.0 mmol/L. By maximizing the sum of sensitivity and specificity, the optimal critical line was determined as 0.69×FPG + 0.14×sAGEP = 7.03, giving a critical sensitivity of 91.2% in detecting 2h-PG ≥11.1 mmol/L, which was significantly higher than that of FPG-HbA1c or HbA1c. The model on FPG-sAGEP improves the efficacy of using FPG alone in detecting diabetes among high-risk Chinese subjects with FPG <7.0 mmol/L, and is worth being promoted for future diabetes screening.