Aims The purpose of this study was to develop a risk score to predict the 4-year risk of diabetes in the middle-aged Korean population. Methods The study sample included 6,342 participants, aged 40–69 years without diabetes at baseline; participants were re-surveyed after 4 years. A logistic regression analysis was used to construct the risk score models to predict the incidence of type 2 diabetes. Personal model was based on simple information such as age, parental or a sibling history of diabetes, smoking status, and BMI. And two clinical models were constructed by adding biochemical tests such as fasting plasma glucose, HDL-cholesterol, triglyceride with/without HbA1C to the personal model. The model accuracy was assessed with area under the receiver operating curves (AROC) and the Hosmer-Lemeshow statistics. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) statistic also calculated to determine the contribution of glycated hemoglobin (HbA1C) to the risk score. Results Of the 6,342 participants, 436 individuals (6.9%) developed diabetes during the follow-up period. Two clinical models that included biochemical tests improved model discrimination capability (AROC, 0.74 and 0.76) when compared with the personal model (AROC, 0.64). But the addition of HbA1C to the clinical model only increased the AROC by only 0.02 despite its high impact on the prediction of diabetes (OR, 2.645). However, the NRI and IDI statistics were significantly improved with the addition of HbA1C. Therefore, a simple risk score was developed to estimate the risk of diabetes based on the second clinical model that included HbA1C. To finalize the model, the internal validity was assessed using a 10-fold cross-validation procedure. Conclusions A risk score derived from simple biochemical examinations including HbA1C can help identify those at a high risk of diabetes in the middle-aged Korean population. Keywords Risk score, Diabetes, glycated hemoglobin, Korean