Purpose: To analyze the relationships between choroidal thickness (CT) and diabetes mellitus (DM), diabetic retinopathy (DR), and DR severity in community residents diagnosed with type 2 DM, and to explore whether CT can improve the discriminatory ability of other risk factors to predict the incidence of DR. Methods: A total of 1,250 type 2 DM residents and 1,027 healthy controls in Xinjing community of Shanghai participated a cross-sectional survey of eye diseases in 2016. CT was measured using swept-source optical coherence tomography. DR was classified according to the 2002 international clinical classification of DR. A total of 537 subjects with type 2 DM without DR at the 2016 survey were followed up in 2018 to investigate the 2-year incidence of DR. Receiver operating characteristic curve analysis was used to test the accuracy of different indicators in predicting the onset of DR. Results: The central CT of the control, no DR, mild non-proliferative DR (NPDR), moderate NPDR, severe NPDR, and PDR groups were 223. 40, 216.22, 213.57, 211.91, 178.47, and 168.15 μm, respectively (p for trend < 0.001), and the average CT (ACT) were 197. 83, 186.94, 182.03, 178.00, 156.91, and 136.72 μm respectively (p for trend < 0.001). Body mass index (BMI), DM duration, fasting blood glucose, glycosylated hemoglobin (HbA1C), and ACT were risk factors for 2-year DR incidence. For the onset of DR, as predicted by ACT, after tenfold cross validation the average area under the curve was 0.55 (p = 0.048). Addition of ACT did not improve the discriminatory ability of DM duration, BMI, glucose and HbA1C on the incidence of DR (Z = 0.48; p = 0.63). Conclusions: As the severity of DR increased, the CT of community type 2 DM patients showed a significant downward trend compared with the healthy controls. Thinner ACT was found to be a risk factor for DR incidence, but it did not improve the discriminatory ability of other risk factors to predict the incidence of DR.