Objective To explore the predictive ability of nomogram for the incidence of type 2 diabetes mellitus (T2DM) in the elderly for 5 and 7 years. Methods This study was based on the data published by the Dryad website of individuals who underwent physical examination in a physical examination center from 2004 to 2015. Finally 712 elderly people without T2DM were enrolled at the baseline, with a follow-up period of 5 and 7 years. According to the T2DM diagnosis at the end of follow-up, all participants were divided into diabetes group (n=679) and non-diabetes group (n=33). The two groups were compared in the demographic and clinical characteristics. Univariate and multivariate Cox regression analysis were used to determine independent risk factors. Based on the findings of Cox regression multivariable analysis, a nomogram was constructed to predict the 5- and 7-year incidence of T2DM in the elderly in China. The receiver operating characteristic (ROC) curve and the concordance index were used to evaluate the differentiation of the model, and the calibration curve was used to evaluate the calibration of the nomogram model. R software was used for statistical analysis, nomogram (4.2.0) was generated based on multivariate prediction model(http://www.r-project.org/). Data comparison between two groups was perfomed using t test, Kruskal-Wallis rank sum test or χ² test depending on data type. Results There were statistically significant differences in fasting blood glucose (FBG), triglyceride (TG), high-density lipoprotein cholesterol, glycosylated hemoglobin Alc (HbA1c) and alanine aminotransferase (ALT) between the two groups (P<0.05 for all). According to Cox regression multivariable analysis of the participants and previous studies, gender, age, body mass index, ALT, TG, HbA1c, FBG were finally included in the nomogram. The area under the ROC curve (AUC) for 5-year was 0.905, and for 7-year was 0.835. The concordance index was 0.850 (95%CI 0.772-0.929), indicating a good discrimination of the model. The calibration curve showed good consistency between the estimated probability and the actual result. Conclusion Our nomogram is a simple and reliable tool for predicting the 5- and 7-year risk of T2DM in the elderly in China. Using this model, early identification of high-risk groups helps to timely intervene and reduce the incidence of T2DM.